Research And Implementation Of De - Fog Algorithm Based On Zynq7000 Platform

In recent years, fog and haze often occur. In the frog, light is affected by suspended particles in the atmosphere which could cause the images obtained by forming apparatus lose its quality, the color of the image is darkened and the contrast reduced. Many negative effects would be brought to daily life and industrial production. So image dehazing is becoming a hotspot in the research of image processing. Meanwhile, considering the application where the dehazing technology is used, like the vehicle-mounted system, monitoring system or the reconnaissance systems, obviously the realization of dehazing in embedded systems should be highly valued.In this paper, firstly a research of all the dehazing algorithm existed is conducted, the basic theories, the ways dehazing realized and the final results were compared. The algorithm of Dark Channel Prior is chosen as the algorithm to be migrated to our hardware platform.Secondly, an analysis of the processing effect and efficiency performance of the chose algorithm is made. As the time of computation is too long and the memory consumption is too large, an improved scheme is proposed, which could significantly reduce the processing time of guided filter and improve the efficiency of the algorithm.Thirdly, the configuration of the hardware environment based on Zynq7000 was realized, including the configuration of operation system and file system. Meanwhile, the the algorithm is implemented using OpenCV.Finally, after a cross compilation, the algorithm is finally optimized and migrated to Zedboard. Function test and real-time test are conducted. The result shows that about a video of resolution of 300*300, our system could process 20 frames per second, which meets the requirement of real-time processing.