Study On Image Defogging Based On Dark Channel Prior Theory

With the deterioration of air quality,there are some disadvantages such as low contrast,low signal to noise ratio,less marginal information,gray concentration distribution and so on under fog and haze weather conditions.It is very important to improve image quality and make it better used in military,traffic,remote sensing,reconnaissance,reconnaissance and other related fields.In this paper,the basic theory of fog imaging and the traditional image defog algorithm are studied,and the deficiency of image defog algorithm based on dark channel theory is analyzed carefully.In view of the shortcoming of introducing the filter,An improved algorithm for obtaining the transmittance is proposed,which can better calculate the transmittance of the fine and ensure the image is not distorted.At the same time,the global atmospheric parameters can be calculated by judging the existence of the sky region and the constraint condition method.Effectively improve the overall defog algorithm processing speed.Aiming at the drawback that the edge of the filtered image is not prominent,this paper proposes an adaptive weighting improvement method based on the optimized Sobel operator for guiding filter.The improved Canny algorithm is applied to the large area sky.The global atmospheric parameters are separated and solved,the global atmospheric parameters are solved by using the constraint condition method under other conditions,the distortion of the highlight region is solved,and the overall effect of image defog is effectively improved.Finally,the image defog effect is evaluated comprehensively from both subjective and objective aspects.The experimental results show that the improved algorithm can improve the efficiency of image defog without degrading,and the improved guide filter can eliminate the noise effectively,and the information Entropy,image contrast,the ratio of visible edges,peak signal to noise ratio,etc.have improved significantly.The improved image defog algorithm is fast,efficient,robust and widely used.It is of great value in the application of license plate recognition,target tracking,crime detection and remote sensing.