A Fast Image Haze Removal Based On Dark Channel Prior

In the weather conditions of fog or haze,the sol particles present in the air can scatter light,which can lead to a serious reduction in the quality of the images obtained.The reason why image defog technology has become a hot issue in the field of image processing and computer vision in the past few years is that the degraded image directly affects the visual effect and application value.With the support of image processing technology,simple and effective realtime defog of the image has a very important significance in improving the robustness(When the parameters in the system change,some other performance characteristics are maintained)and reliability of the visual system.And can improve the image quality,so that it has a better visual effect.Through the observation and statistics of a large number of natural foggy images,the researchers have come to an empirical rule:almost all of the images in some of the local areas,there will always be a color channel with a very low value appearing in some pixels,resulting in a dark channel priors laws and directly recovering a clear natural defog image.The main research content of this paper is to use the simple and effective single image defog algorithm to recover the natural color image.However,in view of its limitations,the computational cost is large and the processing time is long.In order to speed up the processing speed of fog images,this study introduces another processing tool-guided filtering.Through a large number of experiments we found that the effect of guided filtering and soft matting effect is similar.Optimization of transmittance distribution with guided filtering improves the speed of transmittance optimization.This method is much faster than soft matting,and is much more than an order of magnitude.Therefore,in this paper,the guided filtering technology is applied to the image defog process instead of the traditional soft matting algorithm in the dark channel priors algorithm.With this method not only get a better transmission map,the most important thing is to achieve the desired goal,greatly speeding up the processing speed,and improving the speed of 20 times than the use of soft matting method,practicality is enhanced.And in addition to the introduction of the guided filtering to improve the image processing speed,the traditional atmospheric light estimates were modified to solve the color cast problem in the transition area.It is also the highlight of this paper.