Image Enhancement Algorithm Based On Adaptive Fractional Differential

In recent years,the frequency of haze weather has been increasing,and many of the features contained in images taken under this condition have been blurred or some of the key information has been lost,resulting in low image contrast and poor clarity.It caused a lot of inconvenience to the subsequent image processing.Therefore,image enhancement in foggy weather as a preprocessing stage of image processing is playing a decisive role.In this paper,Foggy images enhancement algorithm based on fractional differential is mainly studied.The main contents are as follows:(1)According to the high order approximate formula of Riesz factional differential,the RH operator of traffic image enhancement in fog is established.Considering the relationship between pixels in the image,in order to achieve better texture enhancement effect,the zero elements of the RH operator are replaced,and the improved RH operator—IRH operator is obtained.The experimental results show that the two operators proposed in this paper have a significant enhancement effect on foggy images.Compared with other image enhancement operators,they have obvious advantages in improving contrast and sharpness.(2)To solve the problem of poor defogging when using the fractional-differential algorithm to enhance fog images,the fog image is processed with dark-color prior and Retinex theory to obtain pre-process images firstly.Then,the image threshold segmentation algorithm is used to separate the image into regions with different characteristics.The segmentation function is established by combining different features of the pixels in each region,and the optimal differential order of pixels in each region is determined adaptively.Substituting the order into the RH operator and the IRH operator to perform convolution operation with the image,a new method of self-adaptive fractional derivative with fog image enhancement is formed finally.The experimental results show that the algorithm has obvious enhancement effect in vision,and the values of information entropy and average gradient are higher than those of other algorithms.(3)It is difficult to artificially give the fractional differential order of the image enhancement effect in an image captured in a complex environment,and when the entire image or the segmented region is filtered using the same fractional differential order,it is also difficult to ensure that the information of the remaining pixels can be effectively retained while enhancing the target information.To solve this problem,we use two texture features of image such as gradient and local information entropy to establish a point by point adaptive method,which makes the enhancement purpose more pertinent.The adaptive method is applied to the RH fractional differential operator and the IRH fractional differential operator respectively,and good results have been achieved.