The Research Of Image Haze Removal Algorithm Based On Joint LLSURE Filter

In the poor weather condition caused by haze and fog,the captured pictures have serious quality degradation questions,such as low contrast,color distortion,loss of detail information in the scene,which greatly restricted and influced the full play of the utility of the outdoor computer vision system.Hence,haze removal is an important and widely demanded topic in the fields of computer vision and image enhancement,which has broad application prospects.In the practical application,the speed of image defogging algorithm is a key problem.The single image haze removal using dark channel prior method has greatly attracted researcher's attention because it can restore the haze-free image simply and effectively.However,the algorithm is easy to appear halo effect.To suppress the halo effect,they need to refine the raw transmission map by utilizing soft matting method.But the soft matting method suffers from time consumption and high memory,which cannot meet the requirement of real-time processing.Based on this,this paper analyzed the reason of foggy image degradation and fuzzy mechanism,deeply studied of the basic theory and key technologies of image dehazing processing,and improved the existing image defogging algorithm,introduced new ideas,and made a lot of experiments.Specific research works of this paper can be summarized as follows:(1)According to the soft matting method shortcomings,such as high complexity,large amount of computation.This paper proposed a novel fast raw transmission map refining method based on the joint LLSURE filter.Our improved algorithm with linear time complexity,greatly improves the speed of the method and can meet the requirement of real-time,and better maintains the image edge information,improves the estimate precision of the transmission map,especially for the depth mutation region,at last,it suppresses the halo effect and block effects effectively.(2)The color of dehazed image is easy to appear dark by the existing defogging algorithm,and this phenomenon is called over-dehazing problem.To solve this problem,we deeply analyzed the reason of this phenomenon and studied the atmospheric scattering model,then we defined a imitated-dehazed image based on the atmospheric scattering model,and proposed a fast image dehazing algorithm based on image fusion.That is fuse the defined brighter imitated-dehazed image with the original dehazed image which based on the joint LLSURE filter to solve the over-dehazing problem.The experimental results show that our method has restored the color and contrast of the scene effectively,and still applied to some with heavily haze and serious color distortion images.(3)In the practice,the quality of the kiln flame images are seriously affected by the harsh field environment,and the acquired images are often blurry,which seriously affected the flame image post-processing.According to the existing defogging algorithm invalid for enhancement coal-fired images,this paper presented an image defogging method using the average model-based image fusion,which can improve the estimation accuracy of the flame image restoration.Experimental results show that the restored flame images using the proposed method is more clear and natural,and have good enhancement effects for each region of the flame image.