An Algorithm for Texture Enhancement Based on Fractional Differential Mask Using Adaptive Non-integer Step

Image texture enhancement is an important issue in computer graphics,computer vision and pattern recognition.By analyzing fractional differential principle and characteristics of texture images,a fractional differential mask with self-adaptive non-integer step is proposed to enhance texture images.Firstly,a non-regular autocorrelation mask region is adaptively constructed by taking advantage of the high correlation among textures in images to remove lowly correlated pixels and reduce the interference from noises.Afterwards,breaking through the traditional method in which the step in the numerical calculation has to be taken in unit size,pixels spanning on the arms of autocorrelation region are divided adaptively based on arms' length to further improve the accuracy of fractional differential computation.Finally,local linear modeling is adopted to estimate gray scale value for non-integer pixels.Experimental results show that the proposed algorithm can significantly improve the precision of numerical analysis of fractional differential computation,and thus can further enhance the complex texture details in smooth areas of images.