Research And Implementation On The Technology Of Fast Image-defogging Based On Embedded System

Image in the fog is often degraded by scattering due to atmospheric particles, The distorted and blurred image make it difficult to estimate the objective, and directly limited the outdoor vision systems. Therefore, it's urgent to improve fog-degraded image quality, reduce the effect of the fog.The content mainly consists of two parts: part one is optimizing of defogging algorithm to obtain faster processing speed or better results. the other is combining the characteristics of the TMDXEVM8168 platform to select the appropriate defogging algorithm, and optimize the defogging algorithm.The Adaptive Histogram Equalization image enhancement method is emphatic studied in the enhancement methods. We proposed to add a parameter to solve the problem of excessive enhancement. In the research of dark-channel prior, we proposed a more efficient Minimum filter, and compressing images to reduce the amount of calculation. In the point of filter, using Maximum filter instead of Guided Image Filtering to complete the edge repair work. In addition, we design a Directional Median filter that can filter the noise and preserve edges. In the process of defogging, we used Directional Median filter instead of Minmum filter and Guided Image Filtering to improve the speed of algorithm in dark-channel prior.In the point of Algorithm transplantation, we used some other methods to further reduce the algorithm complexity, such as look-up table, shift operations, to avoid the calculation of floating-point, multiplication and division. we also made full use of the hardware advantages for TMDXEVM8168, so that the processor can focus on algorithm processing. With the help of co processor the speed of platform have dramatically improved, and met the real time requirement.