Proceedings Article | 11 April 2019
KEYWORDS: RGB color model, Image processing, Light sources and illumination, Light emitting diodes, Liquids, Cameras, Interfaces, Reflectivity, Data acquisition, Computing systems
Delivering process of liquid fuels from refinery units into storage plants is no longer using transportation (trains or trucks) because it is so costly. So pipes are used for this delivering process. Densitometer is used to determine the time when valve close or open manually for guiding each fuel to each storage tank. Density parameter is not suitable to be the main parameter because the densities of each oil are almost the same. Therefore it needs to design an innovation of Raspberry-Pi's liquid fuel detection system to improve detection accuracy. in this research, fuel color data acquisition process on a transparent pipe on HSV color space (hue, saturation, value) are done by webcam then data processed by Raspberry - Pi to define the fuel type. Script programming has written in Python 2.7 and image processing is done with OpenCV library. Optimal intensity lighting at 25 lux and minimum at 8 lux. Each fuel and interface can be well known. The values, accuracy, standard deviation of hue and saturation readings respectively for pertamax are 98 - 100, 98.8%, 0,3 and 120 - 140, 98.48%, 3,07, for diesel are 19 - 25, 99.12%, 0,41 and 110 - 130, 97.35%, 3,79, and for kerosene are 92 – 100 , 98.93%, 1,38 and 30 - 45, 92.55%, 4,63. Determining the fuel concentration on the interface can’t be done because there is the same value of HSV bit.
