The Research Of Algorithm Of Haze Removal Based On Dark Channel Prior

In fog weather,the image is poor,reducing the visibility and contrast.Then outdoor visual application system can not run in a reliable way, such as outdoor monitoring,unmanned car,aerial image.So it is necessary to defog.Single Image Haze Removal Using Dark Channel Prior is a novel method, and it provide a new train of thought for our study about haze removal. It has become a classic theory method to fog,and is worth for us to make further studies based on the theory.We find some problems,in the process of research.Such as, airlight is not reasonable in some scene, and the minimum filtering can make dark color expand outward in the process of calculating dark channel, causing the transmittance to expland and be larger, and the edge of dehazing image will have "halo" phenomenon.Bright area is not meet the assumptions of dark channel.The effect of haze removal is not very ideal, and the time complexity is too high. on these issues,this article we find atmosphere light value using hierarchical searching method based on the quad-tree subdivision,and use the theory of morphology to corrode the rough transmittance.Then we use the guided filter to get fine transmittance,and histogram equalization is used to eliminate transmission rate that is close to zero value.At last,this article use model to recruit image.Experiments show that the improved algorithm for haze removal is better, faster and has stronger robustness.