Histogram Based Image Dehaze Algorithm

Over the past decade,image dehazing is a difficult problem,it gets the attention of researchers.Image dehazing algorithms can restore the degraded image,and obtain a clear result.Based on the analysis of previous work,we propose three histogram based image dehazing methods,they are global style transform method,local adaptive style transform method and contrast accumulate histogram equalization,respectively.The experimental results show that the proposed algorithms improve the visual effect,and the computational complexity and computing time of the algorithms are better than the state of the art methods.With the rapid development of science and technology,people have higher and higher requirements on the image or video,not only the high requirements for the pixel of image is needed,but also the natural weather's impact for photo should be small.Especially the image is degraded by smoke,fog and other bad weather conditions.In recent years,researchers have proposed a large number of image dehazing methods and techniques,and some good results are obtained,but most of them are not practical for engineering applications.These methods can be divided into three main categories,the histogram method,the Retinex method,and the method based on the dark channel priori.In this paper,we introduce the representative methods of these three categories,and analysis the advantages and disadvantages of each method.In order to get a fast image dehazing method,this paper introduces the idea of style transform into the image restoration approach.The basic idea is to use a natural image with rich texture and detail as a reference image,and then we transfer the “style”of the degraded image to the “style” of the reference image.In this way,we can effectively improve the visual effects of degraded images,and this approach has fast speed.In this paper,there are two methods for image dehazing based on transform,one is global transforma,the other is local adaptive transform.The former is mainly based on histogram matching,we match the histogram of the degraded image to histogram of the reference image,a gray mapping is constructed to achieve the effect of removing haze.The latter is based on the idea of local adaptive histogram equalization,and we proposes a novel technique of local style transform.The effectiveness of the two algorithms proposed in this paper is verified by several experiments.In addition,the idea of contrast accumulate histogram is introduced into the single image dehaze problem for the first time.The traditional histogram equalization has attracted much attention because of its intuitive implementation quality,high efficiency and monotonicity of luminance mapping function.However,histogram equalization overemphasizes the contrast around pixels that are large in number but of little visual importance.The proposed method adaptively controls contrast gain according to luminance and pixel's potential visual importance.It is found that the details of images in natural scenes are usually hidden in dark areas with obvious local differences.We use the multi-resolution and dark-pass filtering gradient of the images.The potential visual importance was established.Experiments show that our method has strong ability in general image dehazing.