Research On Image Detogging Algorithm Based On Physical Model

In the fog and haze environment,the suspended particles in the atmosphere can make light scattering,which affect the spread of it.In this kind of weather environment,the image quality is low,some image information is lost and the visual effect of image is poor.Image defogging has a high value in the field of computer vision and wide application value in real life.There are some defects and limitations in the existing defogging algorithm,so there is still a lot of room for it to improve.The main contents are follows:Firstly,in this paper,the defogging algorithm was divided into two categories according to different defogging principles: image enhancement algorithm and image restoration algorithm.In this paper,the theoretical model of the image restoration algorithm was introduced in detail,including the incident light attenuation model and the ambient light imaging model.And the principle of fog image degradation was described.Secondly,the traditional defogging algorithm based on dark channel prior was easy to produce the Halo effect,that is,a white halo emerging in the breaking depth of image field.For this reason,a method of scale adaptive dark color defogging was put forward.This method was to use different templates with the minimum filter in the position of the depth ofimage field and the position of not the depth of image field.Then the dark color image and rough transmission image were obtained,and the filter was used to repair the rough transmission image.The simulation results show that the defogging image could be evaluated objectively by using the ratio of new visible edges,the gradient mean and the effective detail intensity,which proved that this algorithm could effectively reduce the Halo effect of defogging image.Thirdly,there were some deficiencies in the atmospheric dissipation function algorithm:firstly,because the tarel algorithm used the median filter to refine the atmospheric dissipation function,so the edge of defogging image maintained not good.Secondly,the defogging effect of tarel algorithm was not ideal for the uneven distribution of fog scene.This paper presented an improved algorithm:Firstly,the histogram equalization was used to preprocess the fog image.Then the atmospheric dissipation function was refined by bilateral filtering.Simulation results show:compared with the traditional tarel algorithm,the algorithm in this paper can get better defoging effect.