Image De-hazing Processing Based On The Dark Channel Prior Algorithm

In foggy weather conditions, serious scattering particles of the incident light and the atmosphere make the image which the imaging system acquired degraded seriously, such as the declining in image contrast, lower resolution, colors dim. These characteristics results in the bad viewing and the worthless using of the image information collection. Therefore, it is necessary to use a fast and effective method for atomizing mist of image processing, and to improve the quality of the original image. Based on the physical model of the atmospheric scattering and dark channel priori principle, this paper analyzes the nature of image degradation deeply, and discusses the shortcomings of the original algorithm from a real-time image processing and visual aspects, then makes corresponding improvements to deal with these deficiencies. This paper is mainly for single degraded image restoration processing.The fixed minimum filter template for images of different size is not conducive to the atmosphere light intensity estimates. An adaptive block template is used in this paper, which can eliminate the blocking effect to a certain extent, also can increase the accuracy of atmospheric intensity estimation; due to the little application of the dark channel priority algorithm in bright areas of the image, which led to the restored image color distortion problem, we introduce a tolerance mechanism to weaken the local area and amend the transmittances of these special areas, ensuring the transmittances distribution can transition smoothly, increasing the occasions of the algorithm using and improving the robustness of the algorithm, transmittance optimization by using soft matting algorithm is with high space and time complexity, which does not meet the real-time, an improved joint bilateral filtering algorithm with linear block acceleration is proposed here to accelerate the implementation of the efficiency of the algorithm, as well as to obtain a better treatment effect. Finally, those analyses of the proposed algorithm have been done by the objective evaluation criteria of the image qualities.Image simulation in this paper bases on MATLAB plat form, the experimental results show that, improvements of the proposed algorithm can recover true color of more atomized scenes image quickly and effectively, improve the contrast, obtain a clear image, and enhance the execution speed of the algorithm greatly, while the time complexity of the algorithm and the image size is linear, basically meet the real-time requirements. It’s possible to ensure the stability and reliability of the imaging system to work out doors. At the same time, the time complexity of the algorithm is only a linear function of the input image size, which can basically meet the real-time requirements.