PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.
This PDF file contains the front matter associated with SPIE Proceedings Volume 12747, including the Title Page, Copyright information, Table of Contents, Introduction, and Conference Committee.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Aiming at the problem of aircraft target detection in complex environment of hyperspectral image, a new target detection method based on space-spectrum combination is proposed. Hyperspectral satellite images are pretreatmented firstly, and the method of band subset global searching is adopted to reduce the spatial dimension, then least squares error algorithm is employed to find the best band combination. To make full use of the spatial information of hyperspectral images, morphological attributes and guiding filter are used in the algorithm. And an abnormal target detection method of hyperspectral images base on attribute filter and guiding filter is proposed to improve the detection accuracy of target detection. Experimental results show that the proposed target detection method based on space-spectrum combination could suppress the false alarm targets in hyperspectral remote sensing images and improve the detection rate of real targets.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The classical traditional dark channel priori algorithm causes colour distortion in bright areas of the sky, snow and other surfaces when processing images, and the overall darkness of the recovered image is complicated, resulting in slow processing speed. In this paper, we propose an adaptive modified dark channel priori defogging algorithm that combines inverse bright channel and dark channel weighting. Firstly, the brightest channel of the RGB channel of the fogged image is inverted, and then the dark channel is weighted and combined with the dark channel to correct the dark channel and adaptively correct the transmittance. Secondly, the mean value of the brightest 0.1% of pixels in the new dark channel is used in estimating atmospheric light values; finally, the transmittance estimation is optimised using multiwindow fitting and guided filtering methods. Experimental results show that the algorithm improves the dark channel priori colour distortion in bright regions, with faster processing and more effective defogging.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The high-order aberration of the human eye is a factor that cannot be ignored in the visual correction. The design of lens that can correct the high-order aberration can reduce the negative impact on the human eye and improve the visual quality. In this paper, the optical design software ZEMAX is used to construct a personalized eye model based on the Liou eye and measured eye data. The detailed optimization process is given when fitting the wavefront aberration, so that the wavefront aberration of the target human eye and the actual human eye tend to be consistent. The constructed personalized eye model has the same optical characteristics as the actual human eye. Based on the personalized eye model, an aspherical lens is designed to correct high-order aberrations. After correction, the high-order aberrations of the target eye are reduced, and the PV of wavefront aberrations is decreased by 52.05%, RMS is reduced by 59.64%. Meanwhile, the MTF of tangential and sagittal direction increased by 180% and 135% at 100 cycles/mm respectively.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
We demonstrated a high-power Tm:YLF slab laser dual-end-pumped by two diode-bar stacks. A continuous-wave (CW) laser output of 235 W at 1.91 μm was obtained under a total incident pump power of 698 W. The corresponding slope efficiency was 42.4%, and the optical conversion efficiency was 33.7%. At the output power of 200 W, the beam quality factors M2 were calculated to be 500 and 2.9 in the horizontal and vertical direction, respectively. In addition, we simulated the temperature distribution of Tm:YLF slab crystal under the maximum output power using the simulation software of COMSOL. Theoretically, we analyzed the focal length of the thermal lens, the spot size of the fundamental mode, and the stability parameter of the cavity. The experimental results and theoretical analysis indicated that the Tm:YLF Innoslab laser was a promising candidate for high-power applications laser technology.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In this paper, a laser diode with a center wavelength of 444nm is used as the pump source. Based on the pre-laser mode selection technology, combined with the F-P etalon technology for auxiliary mode selection, a single-longitudinal-mode pulse laser with a center wavelength of 639.7nm is developed. In the experiment, a stable single-longitudinal-mode laser output under high energy injection is realized by optimizing the parameters of the laser insertion element. When the output power of the pump source is 10.8W and the repetition rate is 20kHz, the maximum single-longitudinal-mode output pulse energy is 10.25μJ, the peak power is 170W, and the pulse width is 77.66ns.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In this paper, the complex energy band of photonic crystals is calculated in weak form, and the diffraction coefficient 𝑛 of the subwavelength grating is obtained, and the chirped subwavelength grating is deterministically designed. For the two polarization states of TE and TM, different subwavelength grating structures are designed, and for the two polarization states, the results of simulation show that the device efficiency reaches 70%(-1.44dB) and 54%(-2.61dB), respectively, and the 3dB bandwidth is greater than 50nm.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In order to improve the measurement accuracy of the barrel angle of the gun, a measurement system of the barrel angle of the large-caliber gun is proposed. The structure and working principle of the test system are introduced. The optical structure of the system is designed. The panoramic image of the barrel is obtained by cone mirror imaging. The rotation angle of the barrel device in the barrel is observed by using the principle of optical over-range angle measurement, so as to solve the problem of measuring the rotation angle of the rifling. The measurement results show that the accuracy of the angle measurement is better than 2′, which fully meets the needs of the actual test. The problem of measuring the angle of the barrel of the large-caliber gun is solved, and the test accuracy is greatly improved.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Aerial imaging equipment has gradually become a crucial information detection equipment in modern war, so the accuracy of information collection has become the main research direction of imaging resolution algorithm. There are many limitations in the process of making and using aerial imaging equipment, which makes it impossible to implement the super resolution algorithm under ideal conditions. Therefore, it is necessary to study the difference resolution algorithm based on the influence of these realistic factors. In this paper, a mathematical model is established for the impact of airborne photoelectric load vibration on the imaging system, and a preliminary analysis is carried out. According to the vibration environment during aviation flight, a super-resolution algorithm based on micro-step is proposed to extract the non-redundant information contained in the multi-frame low-resolution image sequence for fusion processing, and to reconstruct the high-resolution clear image.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In recent years, color constancy has made great progress due to the application of convolutional neural networks (CNNs). However, CNNs cannot extract global features efficiently, and the use of only high-level features for estimation results in a large degree of information loss. In addition, the model based on CNN used for color constancy is not lightweight enough for application. To overcome those problems, a color constancy method based on MobileViT and multi-scale fusion is proposed. The method adopts the improved network structure based on MobileViT as the feature extraction module. It can make the extraction of local features and global representations simultaneously, with a complexity-reduced model. In further, a novel bidirectional multi-scale fusion network with two paths is presented to facilitate feature fusion between different levels. It makes full use of the multi-scale representation with strong semantics. As a result, more accurate illumination estimation is obtained. Experimental results show that compared with other methods, the proposed method has better comprehensive performance. The illumination estimation error of this method reaches the state-of-art level. With the size of parameters 1.24 MB, it is much lighter than mainstream networks.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
This paper introduced a snapshot spectral polarization imaging system, compressive space-dimension dual-coded spectropolarimeter (CSDS). CSDS uses a digital micromirror device (DMD), a micro polarizer array detector (MPA) and a prism-grating-prism (PGP) to reconstruct a spectral linear Stocks 4D data cube with 100 channels and 3 Stocks vectors in a single shot. The reconstructed spectral profile is compared with the measurement results from micro-spectrometer (Ocean Optics STS-VIS). The feasibility and fidelity are verified from the image and spectral reconstruction evaluations. It is demonstrated that the target material can be distinguished by CSDS.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In this study, a three-dimensional edge coupler is designed, which is composed of a semi-elliptical wedged structure and a straight waveguide. 3D FDTD is used to calculate the alignment tolerance and coupling efficiency of the designed 3D edge coupler. Simulation results show the coupling performance of the edge coupler with a semi-elliptical cross section is closely related to the length of the major axis and the minor axis of the ellipse. In 1550 nm band TE and TM modes, the coupling efficiency of semi-elliptical edge couplers at the optimal coupling position is 84.93% and 79.80%, respectively. According to the simulation results, the 1dB coupling tolerance of the coupler is 4-5 µm.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The polarization imaging detector combines polarization imaging with compressed sensing to obtain four polarization angles of information at the same time. By gaining information from a more dimensional dimension, this increases the contrast of the image and improves detection and recognition capabilities. However, the special structure of the polarization imaging detector reduces the imaging resolution. To solve this problem, we propose a combined imaging method that combines polarization imaging and compressed sensing. We compress the polarization information using digital micromirror array encoding, analyze the influence of the DMD on polarization image errors, and reconstruct the high-resolution polarization information image using deep learning networks. Compared to traditional compressed sensing reconstruction methods, our network achieves better reconstruction results and has higher peak signal-to-noise ratio (PSNR).
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
A mask optimization method based on self-calibrated convolutions is proposed in this paper to reduce the imaging distortion caused by optical proximity effect(OPE). The network model was constructed by combining the inverse lithography technology(ILT), and the parameters of the network model were optimized by the dataset for training. The dataset includes the target pattern and the mask optimized by gradient descent method. The network model based on selfcalibrated convolutions can output an optimized mask according to the target pattern, and the optimized mask is passed through the lithography forward model to obtain the exposure pattern. By the simulation experiment, compared with the traditional gradient-based method, proposed method in this paper has high computational efficiency and small error.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The environment in iron and steel smelting furnace is bad, and there are various unknown interference factors. Non-contact temperature measurement using image sensor is very helpful to the complexity of iron and steel smelting. At the same time, it can better study the interference factors to the temperature accuracy of iron and steel smelting furnace. This paper introduces the development of image temperature measurement technology, analyzes the noise in the process of image temperature measurement, and improves the accuracy of subsequent temperature measurement. Then the least square method was used to compare the temperature and gray ratio obtained by color temperature measurement for fitting analysis, and the data with higher accuracy was obtained.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The observation of target trajectory can be realized by the construction of micro-satellite constellations. So with the development of micro-satellite technology, the space-based optical measurement based on micro-satellite constellations becomes more and more popular. For the space-based optical measurement, it is very important to select suitable rendezvous measurement satellite combination in micro-satellite constellations. An optimal election method of rendezvous Measurement satellites in micro-satellite constellations based on GDOP is proposed to select suitable rendezvous measurement satellite. Firstly, the basic GDOP formula is given according to the analysis of pointing measurement error. And then the optimal selection method is designed based on two rendezvous satellite optimal scheme. The simulation results show that the general pointing errors given by the optimal election method are smaller than the errors given by the random method, which improve the accuracy of space-based optical measurement based on microsatellite constellations.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
This paper introduces the design and experiments of a new electrically controlled hyperspectral infrared Fabry-Perot (FP) filter based on Fano resonance. The FP filter reflector material is studied and the DBR layer film system structure is selected for simulation calculations. Ge material is selected for the high refractive index layer and SiO2 material is selected for the low refractive index layer. The dentate structure hyperspectral infrared FP filter is designed by studying the principle of symmetry breaking to produce Fano resonance. The filter transmission spectrum is simulated and calculated using finite element simulation software. Mathematical modeling of the liquid crystal layer is performed to iteratively calculate the variation of the equivalent refractive index of the liquid crystal under different applied voltages. A bilaterally symmetric single-dentete structure filter with liquid crystal filling in the middle was made to test the variation of the transmission spectrum by applying different voltages to verify the correctness of the simulation.The results show that the appearance of a significant Fano resonance peak is observed in the designed device, implying that the design objectives are well achieved
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In this paper, we propose a metamaterial absorber based on a composite structure of graphene and silicon dioxide (SiO2) for the terahertz (THz) region. The structure includes a gold ground and a graphene layer separated by the dielectric layers of silicon dioxide. The absorber is designed to be insensitive to angles and polarization. The full-wave simulation shows that the designed graphene absorber has an absorption rate of over 99.8% at 2.9 THz. The perfect absorption properties can be dynamically modified by adjusting the chemical potential of graphene. Furthermore, the absorber has robust tolerance for oblique incidence and is also insensitive to polarization. Additionally, we conducted an analysis to determine whether the graphene absorber satisfies the impedance matching condition, thereby improving our understanding of the physical mechanism behind the perfect absorption. Because of its simple structure, flexible tunability, and promising performance, the proposed absorber has the potential for applications in developing graphene-based terahertz energy harvesting, modulators, and other areas.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In this research process, in order to improve the ability of interference-based optical image encryption system to resist chosen plaintext attacks, we propose an optical pseudo-image encryption algorithm to enhance the correlation of plaintext. In this work, we introduce an addictional key, which is highly correlated with the secret image. The key can dynamically update the random phase plate and break the linear relationship between the input and output of the optical encryption system, so that the system will not be cracked by chosen plaintext attacks. Furthermore, the number of keys is reduced through improved sine chaos mapping, so that the encryption system can reduce the amount of key storage data while improving the security of the system.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The fiber optic ultrasonic transducer is resistant to electromagnetic interference and is highly multiplexable. Current optical fiber ultrasonic transducers are mainly used in medical imaging and non-destructive testing (NDT). The current ultrasonic transducer excites a weak ultrasonic signal, which limits its application in large area NDT. This work presents a method for the preparation of a large core diameter optical fiber ultrasonic transducer. Through experimental optimization, the transducer was able to excite an ultrasonic signal with a frequency of 0.449 MHz, a pulse width of about 25 μs and a peakto-peak value of 120 mV.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Distributed Acoustic Sensing (DAS), which utilizes phase change sensing, has higher signal-to-noise ratio and sensitivity compared to other optical fiber sensing technologies, and is often used in the field of marine acoustic detection and seismic wave measurement. However, the frequency fluctuation of the output light of the narrow linewidth laser used in the DAS system and the parasitic interference phenomenon in the sensing fiber can generate low-frequency noise, which ultimately leads to very low frequency drift noise in the demodulated signal. Aiming at the problem of frequency drift in long-distance transmission with parasitic interference as the main source, time-varying filtering empirical mode decomposition is used to extract frequency drift for fiber length segmentation, and combined with improved de trend fluctuation analysis to evaluate the intrinsic mode function, correct and fit to solve the frequency drift in long-distance transmission.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Through in-depth research on various noise sources and characteristics of the full link of the CCD camera system, a mathematical model of the CCD camera system noise was established, and the mathematical model of the noise was simulated and analyzed using MATLAB digital simulation software. At the same time, indoor noise testing of the CCD camera was conducted, and the simulation results were basically consistent with the measured results, verifying the correctness of the noise mathematical model. These research conclusions lay a reliable theoretical foundation for the subsequent search for accurate CCD noise suppression methods.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The present study aims to explore the local spatiotemporal consistency of brain connectivity in PD patients. In total, 30 PD patients and 20 health controls (HC) were collected from the Parkinson’s progression markers initiative (PPMI) and their resting state functional MRI images are acquired. FOur-dimensional Consistency of local neural Activities (FOCA) was employed for the first time to explore local brain activity differences between the two groups. To explore the relationship between FOCA values and neuropsychological indices, the Pearson correlation analysis was performed. And receiver operating characteristic (ROC) analysis was carried out to distinguish PD from HC. Compared with HC, PD patients demonstrated decreased FOCA values in the right inferior temporal gyrus (ITG.R), right middle temporal gyrus (MTG.R), right middle occipital gyrus (MOG.R), right superior occipital gyrus, right inferior occipital gyrus, left precentral gyrus (PreCG.L), left postcentral gyrus (PoCG.L) and left rolandic operculum. The FOCA values in MTG.R and ITG.R were positively related to Montreal Cognitive Assessment (MoCA) values. Negative relationships had been found in FOCA values of PreCG.L and MOG.R and Unified Parkinson’s Disease Rating Scale (UPDRS) III scores. What’s more, the ROC analysis showed that the FOCA values of altered brain regions can be used to distinguish PD patients from HC with high performance. PD patients demonstrated disrupted local brain activity in several brain regions. These brain regions are mainly involved in motor, cognitive and emotional functions. The current study provides important insights into understanding the pathophysiological mechanism of PD symptoms.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Raman spectroscopy offers numerous advantages in bacterial identification, including rich molecular information, quick processing, and great sensitivity. However, accurately identifying bacterial species remains challenging due to the similarity of Raman spectra among various species. This paper introduces a method that combines Transformer networks and Raman spectra for the swift and precise identification of pathogenic bacteria. Our lightweight transformer model, called RamanFormer, outperforms conventional convolutional neural network (CNN) models in identification accuracy and model complexity on the Bacteria-ID dataset and Custom-built dataset. RamanFormer has only about 1/35 and 1/184 of the network parameters compared with CNNs. On the Bacteria-ID dataset, RamanFormer reached a state-of-the-art (SOTA) isolate-level accuracy of 87.03%. We also evaluated the model using clinical bacterial isolates and discovered that it had a SOTA of 99.98% identification accuracy in the 8-antibiotic empiric group task using just ten bacterial spectra per patient isolate. Additionally, RamanFormer also achieved 97.32% identification accuracy on the Custombuilt dataset. Our approach is thus capable of quickly and correctly classifying different bacterial pathogens based on the Raman spectra and could be used for additional Raman spectra identification tasks. The code for RamanFormer will be accessed at https://github.com/Bo-Zhou-gogogo/Raman-transformer.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The upper limit of the particle concentration that current optical particle counters (OPC) can detect is generally lower than 100 P/cm3 , while the particle concentration in industry gas is usually on the order of 104 P/cm3 . In the present study, the reasons that the current OPC is limited for high particle concentration were analyzed, and a new high-concentration optical particle counter was developed. The reasons for the low detection limit of the current OPC are as follows: (1) the thickness of the laser is on the order of millimeters; (2) the current signal processing method by detecting the rising edge is sensitive to coverage loss. Based on this, the thickness of the sheet laser was reduced by optimizing the beam shaping components, and the coverage loss of traditional signal processing methods was overcome by counting the pulse peaks. After developing the new OPC, it was used to detect particles with different concentration, and the detecting results were compared with the counting results of a commercial condensation particle counter. The comparison results showed that the developed OPC can accurately detect the particles with the concentration lower than 12000 P/cm3 .
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Traditional focal length measurement is usually limited to small-aperture optical systems, and relies on human visual judgment to have subjective errors. In contrast, the digital optical parameter testing software system is based on modern equipment, such as a large-area CCD camera, a motor-driven precision translation stage, a large-diameter high-quality collimator, and a high-precision three-dimensional turntable. High-precision measuring system. The test system uses the magnification method, based on the sharpness evaluation function, the variable step length climbing focusing search algorithm, and the Mexican hat multi-directional wavelet transform function image edge detection processing algorithm to achieve focal length measurement, improve the accuracy and efficiency of parameter measurement, and also Complete the framework construction for the follow-up optical parameter comprehensive test system.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In the fault detection process for electric multiple units (EMU), the detection of faults in undercarriage components has always been a difficult problem. The emergence of the TEDS (a dynamic fault imaging and detection system for EMUs) has made this process easier, but it still cannot eliminate manual processes. In this paper, a method of fault detection based on the features of the histogram of oriented gradients (HOG) and support vector machine (SVM) is proposed to improve the capability of the TEDS to identify and locate faults in key components.This method can reduce latent incidents and improve efficiency,ensure the safe and efficient operation of the system.First, the object to be detected is located by means of edge detection, and the image of the object is segmented and preprocessed. Then, the corresponding HOG features are extracted and processed, and the object is identified using the SVM. Finally, the aforementioned process is used to conduct experiments on fault detection based on EMU undercarriage images and identify faults such as missing and/or loose bolts. The experimental results show that the fault detection rate of the method proposed in this paper meets the system requirements and the proposed method can effectively detect samples in complex environments
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Newton’s rings pattern is frequently encountered in optical interferometry, and by extracting the phase contained in it, the measured physical parameter information can be obtained. According to the purpose of point-to-point mapping of the image to be analyzed, a method based on UNet++ to extract the phase of Newton’s rings is proposed. Once the network training is completed, the continuous phase including the curvature radius and ring’s center coordinate can be directly predicted from a single Newton’s rings pattern immediately. The relative error of the curvature radius obtained by parameter fitting the phase is less than 0.83%, and the error of ring’s center coordinate is close to 0 pixel. In order to further improve the results of curvature radius estimation, the parameter estimation results obtained by UNet++ are taken as the initial value and corrected by the least-squares fitting method. Experimental results show that for the Newton’s rings pattern containing -5 dB Gaussian noise, the relative error of the corrected curvature radius is no higher than 0.31%.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In this paper, we propose an Att-Siam infrared cluster target detection and tracking algorithm to address the difficult problem of background distinction in the process of long-range detection and tracking of small targets in air clusters. The Att-Siam algorithm first introduces the attention mechanism in the high-dimensional Hilbert spatial data and the original spatial data, and then introduces the attention mechanism in the convolutional channels, the stacked channel attention mechanism, and the spatial attention mechanism. By improving the learning performance of the network structure, the proposed algorithm improves the tracking success rate and accuracy by 32.3% and 20.9%, respectively, when compared with the traditional SiamFC algorithm tested on the spatially weak target IR dataset. The experimental results show that the proposed algorithm can adapt to complex and diverse infrared aerial scenarios and achieve effective and stable real time tracking of small infrared aerial targets.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In the field of flame detection, general target detection algorithms are troubled by low precision and thus difficult to detect small targets. In view of this situation and considering that flame detection requires detection algorithms to be real-time and accurate, a flame detection algorithm based on the attention module is proposed to improve Yolov5 networks. First, in the original network add the lightweight channel attention module ECA, and the network structure is adjusted to increase the feature extraction capability, so as to help the model locate flame features more accurately. Second, the Neck structure is modified by adding a small target detection layer, and in combination with other layers in the feature fusion section, a feature extraction and fusion module is added to improve the small target detection capability and feature fusion capability. Finally, the localization loss function is improved to EIoU Loss to accelerate the model convergence speed while improving the prediction frame regression precision. Experimental results show that the improved flame detection algorithm improves the average detection precision by 7.8% over the Yolov5s model, and the detection speed reaches 39 frame/s at the same time. It is not only real-time and accurate, but also effectively improves the detection capability on small targets.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
An automatic defect detection method for the mobile phone curved glass based on machine vision is proposed and implemented to estimate the plane images, edge images, and R-angle images of the mobile phone curved glass. Moreover, different defect size can be obtained. The experimental results show the consistency with the image measurement instrument, and the common scratches, stains, scratches and bubbles on the curved glass surface of mobile phone can be accurately extracted by the proposed algorithm, with a dimensional accuracy within 20μm.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
A simple structure, high stability, and high speed characterize laser triangulation. Aerospace, civil engineering, and many other applications use laser triangulation inspection methods. In various applications, laser triangulation is affected by a variety of factors. The paper proposes an improved circle fitting method for solving the problem of low accuracy of speckle center detection in laser triangulation systems. There is usually some deviation in the measurement results of circle centers when there are outliers in algebraic or geometric circle fitting methods. The center of the fitted circle is determined by combining edge detection and least squares circle fitting. Laser triangulation is, therefore, more accurate by improving the accuracy of the spot origin.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In order to solve the problem of precise pose measurement by monocular view caused by weather or equipment failure in medium and long-term optical pose measurement in the shooting range, a measurement method of monocular 3 pose angle based on target image matching in linear vector direction for multi linear aircraft targets is proposed in this paper; For missile targets with only single feature of axis vector, the yaw and pitch angles are obtained by optical monocular view pose processing method based on axis vector direction, image length matching; This method not only makes use of the reliability and flexibility of linear feature extraction, but also avoids unnecessary error factors. This method can be applied to the pose processing of axisymmetric and non-axisymmetric rotating targets at the same time, unifies the processing approach for monocular pose processing, and provides a theoretical support for the medium and long-term optical pose measurement in range.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
This experiment uses the laser scattering principle and introduces a CCD camera to study the relationship between scattering offset and liquid concentration, and uses MATLAB interpolation to fit the two relationships. The innovative use of MATLAB for grayscale processing, combined with virtual simulation software for error analysis of the data. The relationship between the liquid concentration and the refractive index of the liquid to light and the refractive streak is finally obtained.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Extreme ultra-violet (EUV) lithography photomask defects are a common problem in the lithography printing process, which has a serious impact on the lithography printing process. Therefore, it is necessary to detect and quickly locate the defect. Many researchers have used image processing and machine learning methods to quickly identify defects in EUV photomasks and subsequently repair them. This paper proposes a detection method based on neural network image segmentation, and we introduce an improved U-Net to predict photomask defects. Our experiments show that the network model has better accuracy. In the process of identifying the defect image, it is in good agreement with the ground truth.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The research of ultrafast dynamic processes needs the support of synchronization technology in the corresponding time scale. To use the THz pulse of China Academy of Engineering Physics Terahertz Free Electron Laser (CTFEL) and an external fs laser to study the ultrafast dynamic process with pump-probe technique needs the femtosecond-level synchronization between the two pulses. This letter introduces previous work and feasibility study of the synchronization method between THz and fs laser and obtains some experimental data for constructing the feedback compensation system. The experiment generates THz pulses by BNA/ZnTe crystal acting as accelerator-based THz light source for simulation, another fs laser pulse (split from the same fs laser, having a delay line to get different timing jitter.) will be modulated and detected to get the characterization using electro-optical sampling (EOS) and sum-frequency generation (SFG) and detection to give the feedback parameter.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The measurement of photoelectric effect and Planck constant is one of the important contents of college physics experiment. The photoelectric effect and Planck constant were measured by experiments, and the volt-ampere characteristics and photoelectric characteristics of the photoelectric tube were analyzed. Matlab was used to write programs to process data and fit images. The results show that the Planck constant and the volt-ampere characteristics of the tube can be well obtained by Matlab, and the error between the measured Planck constant and the recognized value is only 3.72%.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Monitoring wind field changes is of great importance for real-time weather forecasting, military environmental forecasting and space weather situation analysis. Compared with weather balloons, cup wind speed sensors, thermal wind speed sensors, ultrasonic anemometers, wind profilers and other wind measurement tools, laser wind lidar has the significant advantages of high measurement accuracy, meticulous measurement time detection distance. As the research continues, lidar gradually in the civil field as well as the military field has more and more broad application prospects. This paper briefly introduces the working principle of laser wind lidar. Highlights the development history of laser wind lidar. The various types of laser wind lidars are compared, and their respective characteristics are listed. Finally, the development trend and characteristics of laser wind lidar technology are briefly summarized.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Clouds are visible aggregates of small water droplets, supercooled water droplets, ice crystals or their mixtures suspended in the atmosphere; sometimes they also contain some larger raindrops, ice particles and snow crystals whose bottoms do not touch the ground[1]. The observation of clouds is an important part of meteorological observation, and the accurate acquisition of cloud information is of great importance for climate research, weather forecasting, and water resources management, among many other fields. Cloud amount, cloud type and cloud base height are the three elements of cloud observation in meteorological operations, and are also important statistics when analyzing cloud data[2]. Currently, only the cloud height measurement has been achieved, while there are no mature technologies and instruments for observing cloud type and cloud amount, and is still achieved through manual observation. In this paper, the ground-based observation technologies of cloud type and amount have been summarized, and the research status of cloud type identification and cloud amount observation have been analyzed and compared. On the basis of the image processing technologies, the development trend of cloud type and amount observation technologies are prospected by considering the number, quality, and feature extraction methods of samples.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
To identify damaged areas accurately and efficiently in quay crane metal structures and extract their morphology features, a method combines deep learning-based semantic segmentation and morphology feature extraction based on image processing technology is proposed. MobileNet V3 is improved with scSE attention modules and dilated convolutions and used as the backbone network of Deeplab V3+. To adapt to the task of quay crane metal structure damage detection, the output stride in Deeplab V3+ is modified. The class-based Focal loss function is used in training to address imbalanced samples and poor crack segmentation. Experimental results show that the proposed method achieves a mean pixel accuracy of 0.813, a mean intersection over union of 0.787, and a frame rate of 32.2 frames/s. Furthermore, the use of the class-based Focal loss function improved the issue of imbalanced samples and enhanced the performance of crack segmentation. Image processing techniques are used to extract the morphology features of cracks' length, average width, and corrosion area based on the segmentation results. The proposed method accurately segments the damaged areas and extracts the morphology features of the damage, demonstrating its potential for practical applications in quay crane metal structured inspection.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Surface enhanced Raman spectroscopy (SERS) is a novel non-destructive, accurate and rapid detection technology. However, the traditional SERS substrate has the problems of high cost,low efficiency and non-reusability, which limits its practical application. In recent years, researchers are increasingly interested in combining semiconductor materials with noble metal materials. In this report, Au-decorated ZnO nanorods are fabricated on copper sheet (Au/ZNRs/Cu) as a self-cleaning SERS substrate. The ZnO nanorods are prepared by hydrothermal approach. Then the obtained nanostructure is decorated with Au NPs through sputtering. The prepared substrate can detect R6G with a detection limit of 10-7 M. After 30 min of UV irradiation, 85% of R6G molecules is removed.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Newton's ring pattern is an typical interference fringe often encountered in optical measurements. The physical parameters, such as curvature radius, the ring's center can be estimated by analyzing it. Newton ring formed by spherical interference is two-dimensional Chirp signal, and its chirp parameters are almost impossible to be integers. However, the strict constraints of discrete Chirp Fourier transform (DCFT) and the existing discrete Modified Chirp Fourier transform (MDCFT) can only estimate the integer value part of parameter, which has a large recognition error. An improved MDCFT algorithm is proposed to further estimate the non-integer value part by using the object offset principle, thereby improving the estimation accuracy of Newton ring parameters. Experimental results verify the effectiveness and correctness of the proposed method.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In moving object detection research, the result of traditional mixed Gaussian model algorithm are often splited, and effect the results of segmentation and tracking.In our research, the image is detected by the super pixels , and then we use the super pixels as the unit for the following detection, which can overcome the splited objects and get the better effect then the pixel is used as the unit for detection.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In-line digital holography is wildly used to measure object field. While it might on one hand is outstanding in non-contact, simple structure and high resolution, on the other hand its zero-order terms, real and conjugated images are aliased together, also the information of real image is not legible. Compressive holography applies optimization algorithms to eliminate the interference from in-line digital holography imaging, its algorithm model requires zero-padding to avoid cyclic convolution artifacts. Based on the equally divided block compression holography model, this paper illustrates a combination of zeropadding and data-padding, that fulfils a certain amount of redundant information sub-hologram based on the equidistant hologram. The redundant information protects the boundary in the compression holographic model operation, and achieves the high-resolution reconstruction of the in-line hologram. In addition, the method can reconstruct in-line hologram, plus the lateral resolution can reach 1 μm in bubble sample experiment. This is a thirtyfold increase compared to the national standard GBT 15489.2-1995. The relative error of bubble diameter measurement is only 0.48%, which is substantially lower than the traditional techniques.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In this paper, we propose a metal corrosion monitoring method based on highly dense strain field reconstruction to quantify the degree of metal corrosion in real-time. Discrete strain changes caused by corrosion can be acquired by dense grating array fiber optic sensor networks and Optical Frequency Domain Reflectometry (OFDR) technology. Then, they are reconstructed as a non-uniform strain field corresponding to corrosion damage using an image size transformation scheme, hence effectively assessing the relationship between metal corrosion and strain field changes. In the non-uniform strain field, the strain at the boundary of the corrosion region appears a sudden downward trend, and the gradient amplitude field of the strain change can reflect this change feature. The corrosion boundary is extracted by the gradient amplitude field and corrosion extent is assessed. The experimental results show that the proposed scheme can achieve a high-precision metal corrosion location with an accuracy rate over 96% on a corroded cantilever metal sheet, proving the feasibility of the metal corrosion monitoring method based on the reconstruction of highly dense strain fields.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
As one of the common road diseases, the accurate detection of potholes during inspections can help to make timely maintenance measures, which will greatly save road maintenance costs and reduce the incidence of traffic accidents. In order to improve the accuracy and timeliness of pothole detection and to facilitate the development of disease maintenance, the YOLOv7 model is improved. Firstly, the Efficient convolution operator (DSConv) is introduced to reconstruct the Backbone and Head parts of YOLOv7 to reduce the computational effort of the original YOLOv7 model and improve the detection speed of the model. Secondly, the SE attention mechanism is incorporated into the model to improve the model's ability to extract features from potholes. Finally, the latest v3 version of Wise-IoU is introduced as the loss function of the improved model to reduce the impact caused by sample annotation. The accuracy and mAP of the improved model improved by 3.04% and 1.34% respectively compared to the original YOLOv7 model, and the number of FLOPS of the model decreased from 105.1 to 48.8. The results show that the proposed improved method can effectively improve the speed and accuracy of the YOLOv7 model in detecting potholes, and is advanced compared to the current mainstream target detection algorithms.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
An X-ray security image detection model incorporating a multi-scale fusion module is proposed to address the problem of low accuracy in detecting threat objects in X-ray images against complex backgrounds. The model adds a multichannel fusion convolution block after the Neck layer to perform adaptive feature fusion and refinement on the input image, effectively improve the description of global information and boundary attributes of x-ray threat objects to improve the precision of detecting and identifying threat objects. SIoU is chosen to replace CIoU as the loss function of border regression, which redefines the penalty index and reduces the total degrees of freedom of loss to achieve the high accuracy localization. The model can effectively detect five different categories of dangerous goods on the Tianchi dataset, and the mAP value for dangerous goods detection is 92.7%, which is 2.1% higher than YOLOv5s, can satisfy the real-time recognition and detection requirements with high accuracy, good robustness and speed.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
A multi-beam interference theory called internal-external cavity Vernier effect in cascaded Fabry–Pérot interferometer (FPI) was proposed and analyzed. Through that theory, simultaneous measurement of relative humidity (RH) and temperature could be achieved. The cascaded FPI consists of an air cavity and a chitosan cavity. The envelope of the spectrum of the proposed sensor can restore the information of the chitosan cavity, and the high frequency fringe is the sum of the information of the two cavities. By tracing the wavelength shifts of the envelope and high-frequency fringes, a sensitivity coefficient (SC) matrix of the relatively humidity (RH) and temperature could be obtained. The experimental RH sensitivities for the envelope and high-frequency fringe were 1.599 nm/%RH and 0.071 nm/%RH, respectively. Temperature sensitivities of -0.045 nm/℃and 0.035/℃ were obtained for the envelope and high-frequency fringe, respectively. The crosstalk between the RH and temperature was eliminated by the SC. This Vernier effect provides a dual-parameter measurement function and explains the multi-beam interference in cascaded FPI.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Point projection for Non-Uniform Rational B-Splines (NURBS) surfaces is a fundamental operator in Computer-Aided modeling (CAD) modeling. This operator takes as input a query point and a NURBS surface in ℝ3, and outputs the UV parameter values whose corresponding 3D point (one on the surface) has the minimum distance between the query point and a variable point on the surface. Existing projection methods consistently employ an iterative searching strategy, which suffers from efficiency issues, especially for tasks involving enormous query points such as 3D CAD model quality testing. This paper proposes a parallel, subdivision-based strategy to increase query speed. It uses geometric subdivision to quickly cull out potential parameter regions where a query point’s corresponding UV parameters reside, then utilizes the Gauss- Newton method to quickly march to the precise UV values. All steps are done completely on GPU: the geometric subdivision is shared across all query points, and the Gauss-Newton marching is done in parallel as well. Experimental results show that a significant increase of at least 11x can be attained using the proposed method.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Research and development of automatic fruit picking robot is an effective way to solve the low efficiency of manual picking and reduce the cost of picking. Picking robots usually need to use the vision system to obtain scene information, and use image processing methods to detect and locate fruit. In this paper, the RGBD camera based on TOF technology is used to collect data. At first, a method to segment the cherry tomato region on the RGBD image by combining color and depth information is proposed, and then according to the segmented cherry tomato region, the point cloud normal vector is incorporated into the RGB image for data preprocessing. Label the preprocessed data set to make data set, train and test on multiple deep learning algorithms, and the experiment shows that the accuracy has been improved to a certain extent. And improve the input and output information post-training based on the yolov7 algorithm. The experimental results show that the improved algorithm map (0.5-0.95) improves by 5.1%, which can meet the detection requirements based on the picking robot
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Computer-Generated Hologram (CGH) can be used to detect aspheric surface with high accuracy. In order to improve the encoding efficiency of CGH, this paper proposes an encoding method to describe the engraved fringes in segments using the arc as the primitive. Firstly, the phase contour boundary is obtained, and then the boundary is discretized by Newton iteration method. Finally, the discrete points are encoded with the circular arc as the primitive based on the minimum rms criterion of residual error, and the engraved fringes are obtained. In this paper, the CGH is designed, encoded and fabricated for an aspheric surface. The encoding file is only 7M under the premise of encoding accuracy better than λ /100 rms, which proves that this method can greatly improve the encoding efficiency compared with linear encoding. Finally, all errors of CGH are analyzed and summarized, and the root-mean-square error of all errors was 0.00283λ ,which proves the feasibility of the encoding method in this paper.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In this paper, the compression algorithm recommended by the Consultative Committee on Space Data System (CCSDS) 123 standard is optimized according to the spectral correlation of snapshot mosaic hyperspectral images (SSM HSI) to achieve a higher compression ratio. A novel inter-spectral processing module is added to the predictor to calculate the inter-spectral correlation between each spectral band of the hyperspectral image. The required one-dimensional spectral neighborhoods for prediction are selected based on the highest to lowest correlation order, which allow more accurate prediction of the current sample. This approach improves the efficiency of compression. The optimization was performed for Fast Lossless (FL) predictor with sample adaptive encoder recommended by CCSDS-123.0-B-1 standard, FL Extended (FLEX) predictor with sample adaptive encoder, and hybrid encoder recommended by CCSDS-123.0-B-2 standard, respectively, to compare the compression ratio before and after optimization. The results showed that the compression performance was improved by 0.015%-3.1%, 0.01%-2.29% and 0.01%-1.95% respectively. Further optimization with larger compression ratio could be realized by parameters adjustment of the algorithm proposed in this paper.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Due to the low contrast and color distortion of underwater images caused by the absorption and scattering of light by water, this paper proposes an underwater image enhancement strategy based on the attention mechanism of pyramids. Based on FUnIE-GAN, the feature extraction module uses MobileNet to replace the VGG16 model in the original U-Net structure, which reduces the number of network model parameters and accelerates the inference speed of the network model. Furthermore, the pyramid attention module is introduced into the generative network. The multi-scale pyramid feature and attention mechanism collection is beneficial to enhance the network feature extraction ability and improve the model performance. Experiments are carried out on EUVP dataset. The results show that the underwater images enhanced by our method are better in terms of subjective and objective aspects and have improved sharpness, color correction and contrast. The average values of peak signal-to-noise ratio and structural similarity were 21.398 and 0.742, respectively that is better than the other comparison methods.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In response to the traditional detection methods of not detecting the target, not seeing it, blurring the outline and details of the image, etc., using a combination of infrared and polarization detection, the infrared image information and the polarization image information are decoded to solve the problem of not being detected or seen in various environments. For the target local feature extraction process of large amounts of data, slow extraction speed, and other problems, an improved SIFT algorithm for local feature extraction of polarized images with deep learning is proposed. Experimental results show that the modified algorithm combines the advantages of polarimetric imaging and deep learning to achieve fast feature extraction of targets in simple or complex backgrounds. The algorithm improves the speed of local feature extraction of polarized images by 3.6% and the accuracy of extraction by 2.8%. The algorithm provides the theoretical basis for target classification, identification, and tracking techniques.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Aiming to solve the problems of high false alarm rate and single oil-type identification in the existing sea surface oil spill detection techniques, based on the Fresnel theory, a polarization degree model of rough sea surface oil spill under different azimuth and zenith angles is established, and its effectiveness is analyzed along with the influence of polarization distribution. We conducted a visible light polarization bidirectional reflectance distribution test to obtain the visible light polarization degree images of four typical marine oil spills, including those involving heavy oil, crude oil, gasoline, and diesel, at different observation azimuths and zenith angles, and compared them by extracting image gray data. By analyzing the difference in polarization degrees of different oil types, the experiment proved the correctness of the theoretical model. The test results indicate that the visible light polarization characteristics of different oil spills are evidently different. The polarization degree information is influenced by the observation angle. It is easier to distinguish oil types inside view than that in overhead view. The use of visible light polarization to distinguish oil species is a useful supplement to traditional oil spill detection methods and is of great significance to improving the ability of marine pollution control.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The theory of compressed sensing shows that the original signal can be recovered by low sampling rate, so it is often used in the field of optical imaging. To solve the problem of excessive amount of image data and large computational burden, a new method based on column blocking and mixed blocking method is proposed in this paper. Simulation experiments and comparative analysis show that the proposed column blocking method has improved the quality of image reconstruction to a certain extent, while the mixed blocking method has significantly improved the speed and quality of image reconstruction.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Polarization imaging technology integrates the spatial and polarization information of the target scene, which can provide high-dimensional light field information to improve the ability of object detection and recognition. The polarization states of natural scenes can be characterized by the Stokes vector ( S0 , S1 , S2 ), degree of polarization ( DoP ) and angle of polarization ( AoP ). In order to better understand and utilize the polarization characteristics, the observers need to recognize the feature maps of different polarization parameters. These images are sometimes hard to distinguish with naked eyes, especially for S1 and S2 images due to their similarity. This paper proposes a polarization image recognition method based on the cascade deep learning approach, which can improve the discrimination between S1 and S2 images, and achieve preferable recognition accuracy for different kinds of polarization images. We use two ResNet-50 networks successively to classify the polarization images. Firstly, a ResNet-50 network is used to recognize S0 , S12 , DoP and AoP images, where S12 means the union set of S1 and S2 images. Next, the Sobel operation is applied to enhance the discriminat1on of polarization characteristics between S1 and S2 images. After that, the second ResNet-50 network is used to separate the images of S1 and S2 . It shows that the proposed method outperforms some other comparative methods in terms of recognition accuracy.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
In view of the phenomenon of detail loss and edge blurring in color images collected underwater, noise reduction is mostly ignored in underwater image enhancement research. In this paper, an improved image enhancement method based on Retinex algorithm is proposed. First, the original underwater image is transferred from RGB channels to HSI color space, and the saturation component is adaptively linearly stretched by gamma correction, The improved Retinex decomposition network based on homomorphic filtering is used to decompose and enhance the Lightness component separately. Eventually, the fused image is transferred from HSI to RGB color space, and the processed image is output. According to the experimental results, this algorithm can not only eliminate the halo artifacts, but also improve the brightness and contrast of the image, restore the original details of the image, and effectively remove the image noise. It greatly improves the visual effect and objective evaluation results of images.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Feature fusion is a key problem in 3D object tracking, especially in sparse and disordered point clouds scenes. The purpose of feature fusion is to achieve the communication and integration of template features and search features, so as to obtain the fusion features with object-specific information. However, most pervious Transformer-based methods use the SelfAttention Module(SAM) and Cross-Attention Module(CAM) to conduct attention operations progressively in two steps, which is not conducive to focus on the discriminative features from the beginning. Benefiting from the flexibility of attention operations, we propose a Feature-Concatenated Attention Module (FCAM) for ego-feature enhancement and cross-feature augment at the same time. Based on FCAM, we propose a Feature-Concatenated Transformer (FCT) framework to explore more effective 3D object tracking method. This scheme is more useful to achieve deeper integration and extensive communication between template and search features, which makes feature fusion more efficient. In order to verify the performance of the proposed framework, we carried out experimental verification on KITTI datasets. The results of the experiment indicate that our method is superior to the existing schemes in tracking success and accuracy for different object categories.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
An improved image defogging method is proposed to address the problem that the use of dark channel a priori principle for image defogging is likely to cause distortion in the sky region and the overall darkness of the recovered image. In this paper, based on the idea of sky segmentation, we use Gaussian mixture model and EM algorithm to divide the fogged image containing sky region into sky region and non-sky region, and take the mean value of the pixel value of the sky part as the atmospheric light value after getting the sky region. The transmittance is estimated by calculating the RAS color channel to eliminate the effect of atmospheric scattering, based on which the transmittance compensation function is constructed, and the transmittance is corrected by using the constructed transmittance compensation function to reduce the transmittance estimation error. The experimental results show that the algorithm proposed in this paper has a good defogging effect.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Due to the low foreground brightness and distorted background brightness of traffic backlight images, a technique of fusing LIME and improved histogram equalization is proposed to eliminate the backlight phenomenon in images. Firstly, the input image is segmented into two parts, foreground and background, using the maximum interclass variance method. Then LIME method is used globally for the backlit image to enhance the foreground brightness while maintaining the color distortion, and only the foreground part of the processed image is retained. Then the pixel value distribution of the background part is calculated separately, and the global histogram equalization results on the three RGB channels are mapped one by one to the corresponding limited interval, which improves the contrast of the background. Finally, the Canny operator is used to detect the black edges at the front background stitching, and three adaptive filtering templates are generated based on the black edges to perform step-by-step mean filtering on the black edges, eliminating the black edges and improving the visual quality of the image. The average metrics of the proposed method on the laboratory selfconstructed CHD_B dataset are, respectively, NIQE of 5.37, BRISQUE of 47.74, average gradient of 0.79, information entropy of 6.61, and average running time of 11.38s, which are synthetically better than the current backlight image enhancement methods and have better visual quality and visibility
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
This thesis is based on the SigmaStar SSD268G core board, the SONY IMX585 image sensor chip and Linux system to design a high performance and portable UVC high resolution microscope camera. The camera can simultaneously output real-time YUV format data based on USB3.0 interface while realizing real-time image output through HDMI interface and it can also synchronously output high-quality encoded compressed data in H.265 format. The camera can meet the user's requirements for real-time uncompressed data and high frame rate respectively. In order to improve the confidentiality and convenience of the camera control, this paper designs a UVC private protocol to achieve the control of ISP adjustment. The camera designed in this paper can be widely used in industrial detection, biomedical, education and other fields.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Medical images serve as an important carrier and auxiliary tool for modern medical workers to analyze and study health conditions. However, the fact that many medical images have unclear features, and low definition makes it impossible to provide the necessary support for subsequent detection. Therefore, image enhancement based on frame accumulation and centroid registration is put forward in this paper. First of all, in this paper, the experimental equipment for collecting simulated breast images is briefly introduced, followed by the analysis of the collected image data. It has been found that the image presents insufficient feature information due to the scattering characteristic. According to this feature, it is suggested to use the centroid of the image to match the micro-motion images between the frame sequences, and then employ the frame accumulation technique to synthesize, and linearly stretch the frame-accumulated image to improve the imaging clarity. The results show that the information entropy value of the image has been increased from 4.5780 to 6.2734 with an increase of 1.6954. To further prove the effectiveness of the algorithm, comparisons with other algorithms from both subjective and objective perspectives are conducted in this paper to show the superiority of the algorithm.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Spectral computed tomography (spectral CT) is an emerging imaging technology that is capable of distinguishing material properties. However, the difficulty of decomposition process is intensified by the nonlinearity of the measurements and ill-conditioned problem, particularly when the number of materials and energies do not match. Therefore, the key issue in spectral CT is to design an improved algorithm in the accurate material decomposition. This study proposes a one-step multi-material algorithm that combines a statistical reconstruction model with a gradient-sparsity based prior. In our approach, the elimination based method under volume conservation constrained condition is designed for the inconsistent scanning of number of materials and energies. Newton descent method is adopted to efficiently solve the optimization problem based on a simple surrogate function. Through simulated experiments, the proposed method achieves a significantly higher peak signal-to-noise ratio (PSNR) compared to other algorithms, with an increase of approximately 23.988 dB and 23.462 dB. Numerical experiments have confirmed the efficiency of our proposed method in reconstructing the material distributions while reducing noise compared to state-of-the-art methods.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Due to the great influence of fog on life and production, this paper uses Stokes principle in polarization imaging combined with the physical model of fog imaging to design and implement an infrared camera fog removal method based on polarization imaging. In this paper, the defogging method of infrared camera based on polarization imaging is introduced in detail, and the sensor is verified by outdoor experiments. The experimental results show that the overall defogging effect of the proposed defogging method is good, with more details of the target scene preserved and strong contrast, which makes the restoration degree of the target scene higher than that of several classical defogging methods.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Although the increase of network depth and width can increase the accuracy of recognition, the number of parameters and computation are often large, which is not suitable for mobile device applications. To solve this problem, two lightweight CNN models are constructed to improve the feature extraction and feature reuse capability of the network by improving the Fire module of SqueezeNet network, adding a spatial attention mechanism and adding a dense connection module in the deeper layer of the network. The improved models achieved 89.60% and 94.37% recognition accuracy by training on the constructed apple disease leaf dataset, which is 2.29% and 7.75% higher than the original network, while the number of parameters of the network is only 0.9M and 2.5M. The experimental results show that the improved network achieves higher recognition accuracy while keeping the lightweight model.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Facial expression classification is widely used in various research fields. This paper putts forward to an algorithm based on deep learning to solve the problem. Firstly, it segments the facial region and transforms it to gray image, then constructs a convolutional neural network which includes convolution layers and max-pooling layers and full connection layers. The network can classify the facial expression to two categories which are happiness and sadness. Experiments are used to test and verify the network. And it has achieved good results
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Optical imaging is a crucial and direct technique to obtain information, and plays a pivotal role in science and technologies. However, in practical environments, the presence of fog can degrade quality of image, to even invisible. To reconstruct clear images in the presence of fog remains a daunting task in the field of optical imaging. Ghost imaging can suppress noise that is uncorrelated to illumination. This significantly enhances imaging robustness under noisy environment. In this paper, we built a time-gated computational ghost imaging system and evaluate the capability to reconstruct image under weak echoes. Furthermore, we use the imaging system to obtain clear pictures in foggy environments.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
As a new low noise CCD signal processing technology, digital correlation double sampling noise reduction technology is widely used in the field of faint astronomical target detection. In this paper, various noise sources of CCD video processing circuit are studied, the characteristics of various noises are analyzed deeply, the mathematical model of CCD circuit noise is established, and a noise reduction method based on digital correlation double sampling technology is proposed. Through simulation modeling of CCD noise signal, the simulation results show that the digital correlation double sampling method using double Gaussian filter function has a good effect on noise suppression.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
When the patrol robot recognizes the instrument under strong light, the collected instrument image has serious exposure, details loss and other phenomena. Aiming at the above problems, an improved CycleGAN method for enhancing the instrument image in the substation under strong light is proposed. First of all, for the generator, the CSD network and the self-attention mechanism ACmix module are improved to improve the lighting processing effect. Secondly, the idea of Patch-GAN is adopted in the discriminator, and the feature graph is mapped to an N×N matrix at the end to improve the processing ability of the details. The experimental results show that the peak signal-to-noise ratio of the image is increased from 15dB to 20dB after the above improvement; The index of structural similarity increased from 0.5 to 0.8; The recognition accuracy after enhancement has also been improved by 6%. In conclusion, the method in this paper can eliminate the influence of illumination on instrument recognition, and has strong robustness.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Segment the rocket target from the live image has a wide range of application scenarios in the space launch. Aiming at the problem of determining the optimal threshold in the traditional image segmentation, this paper proposed a method to determine the optimal threshold and then carry out image segmentation based on genetic algorithm. And the image segmentation result obviously enhanced, but there are many judgment and large amount of code, which need to be further optimized in the follow-up research.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
A spatial non-cooperative target point cloud feature extraction method based on deep learning is designed for the characteristics of weak geometric information of noncooperative targets. A convolutional neural network model is used to learn and fully exploit the non-cooperative target surface geometric information in a data-driven manner to obtain a local feature descriptor NCPD (Non-Cooperative target Point Describer), which solves the existing problem of insufficient descriptiveness of manual local geometric features. The matching test results in the non-cooperative target point cloud sequence show that the proposed local descriptor outperforms FPFH, SHOT, VoxelNet, PointNet and other methods in terms of descriptiveness and robustness.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Lung cancer is currently one of the malignant tumors that poses the biggest threat to health and life, as both its morbidity and death are increasing globally. The deep learning model has limited impact on the supplementary diagnostic accuracy of common medical samples when the morphological traits are unclear. More spectrum information may be found in the intracellular fluorescent fingerprint data from hyperspectral imaging, creating a novel sample type for tasks involving lung cancer categorization. This study examines the classification challenge of benign and malignant lung cancer using a variety of deep learning models. According to the experimental findings, hyperspectral fluorescence pictures may more clearly distinguish between benign and malignant lung cancer features. In one-dimensional data samples, convolutional neural networks perform better than random forests, but in two-dimensional data samples, they perform worse than residual network models. The 50-layer residual network model, with an accuracy of 0.98, has the greatest classification performance among the three deep residual network models. Hyperspectral fluorescence pictures have been proven to have improved outcomes in the detection of benign and malignant lung cancer through the research, which can better suit clinical needs and aid physicians in making clinical judgments.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
The speed measurement of high-speed moving targets has extremely important significance in military fields, unmanned driving, space monitoring, and other fields. In digital image processing technology, the target coordinate system is often established through video frames for speed measurement. However, in some strong light environments, due to the strong ambient light, the reflected light of the moving target may be "submerged" in the ambient light, and the reflected light cannot be recognized by the detector in the end. Therefore, the motion trajectory of the moving target cannot be identified. For high-speed moving objects, using the backlight method is an effective method for measuring the trajectory of moving targets, but in strong light environments, the "background" light will also be unrecognizable.
This article proposes a new method for accurately measuring the speed of high-speed moving targets in bright light environments. Through spectral analysis of bright light environment, laser is selected as the background light. Laser has the characteristics of high power, strong energy, good penetration, and single wavelength. Its energy in the laser band (1064nm) is much higher than other wavelengths in the bright light environment, thus ensuring the stability of the light source. At the same time, the detection device adopts a band-pass Filter design to attenuate the energy outside the specific laser band and only detect the "background" light. And dynamically adaptively adjust the video frame number of highspeed moving target acquisition devices to improve speed measurement accuracy and reduce bandwidth pressure. The experimental results show that the method proposed in this paper can accurately and efficiently identify and measure the speed of high-speed targets in strong light environments.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
Food image security identification is a research hotspot in the fields of computer vision, data mining, food science and technology, etc. In order to solve the problem that the recognition rate of traditional food image security identification methods is not high because of the small difference between classes and large difference within classes, we propose an enhanced ResNet network food image security identification method. This method combines asymmetric convolution to enhance the learning of local skeleton information, and embeds the attention module shared by deep and shallow layers to solve the undifferentiated feature extraction of the whole image information, which improves the efficiency of feature extraction from local to global. Through a large number of experiments on food image data sets, the results show that the proposed algorithm makes the recognition accuracy of food images reach 85.26% and 96.21%, and it is 100% higher than the popular RESNET 101, RESNET-18 and RESNET-34 model methods. It further shows that the food image security recognition method in this paper will have a good application in small and medium-sized food image recognition systems.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.