Towards the situation that the traditional fractional difference always has the same order in various regions of image for image processing, this paper presents a novel approach that the fractional order is automatically adjusted according to the gradient of every image pixels. The main contribution of this work is that Otsu algorithm is used to design the adaptive image enhancement operator, and calculate the maximum gradient of eight directions for every pixls of image, finally, the optimal threshold of imagecan be calculated by the novel method, which can clearly distinguish the strong edge pixels of image, the texture of image and the smooth region of image.
Information entropy and average gradient are used for quantitative analysis and experimental verification. Compared with traditional fractional differential method, the results show that this presented method can not only obtain more edge information, but also reserve texture. It provides an effective method of image enhancement.