|
An adaptive infrared image enhancement algorithm for low contrast is proposed in this paper, to deal with the problem that conventional image enhancement algorithm is
not able to effective identify the interesting region when dynamic range is large in image.
This algorithm begin with the human visual perception characteristics, take account of
the global adaptive image enhancement and local feature boost, not only the contrast of
image is raised, but also the texture of picture is more distinct. Firstly, the global image dynamic range is adjusted from the overall, the dynamic range of original image and
display grayscale form corresponding relationship, the gray scale of bright object is
raised and the the gray scale of dark target is reduced at the same time, to improve the overall image contrast. Secondly, the corresponding filtering algorithm is used on the
current point and its neighborhood pixels to extract image texture information, to adjust
the brightness of the current point in order to enhance the local contrast of the image. The
algorithm overcomes the default that the outline is easy to vague in traditional edge detection algorithm, and ensure the distinctness of texture detail in image enhancement.
Lastly, we normalize the global luminance adjustment image and the local brightness
adjustment image, to ensure a smooth transition of image details. A lot of experiments is
made to compare the algorithm proposed in this paper with other convention image enhancement algorithm, and two groups of vague IR image are taken in experiment.
Experiments show that: the contrast ratio of the picture is boosted after handled by histogram equalization algorithm, but the detail of the picture is not clear, the detail of the picture can be distinguished after handled by the Retinex algorithm. The image after deal with by self-adaptive enhancement algorithm proposed in this paper becomes clear in
details, and the image contrast is markedly improved in compared with Retinex algorithm.
|
