Abstract: An improved scheme is given out, which can realize the nonlinear retaining of texture detail by using self-adaptive Fractional-order calculus algarithm to conduct wavelet coefficients, in order to solve the problem of losing texture detail existing in the process of image enhancement when the traditional wavelet algorithm is used. First of all, the gray-level similarity weighting factor of each pixel in the original image is calculated, and the original image is decomposed into n-tier by two-dimensional wavelet. The decomposed low frequency, horizontal, vertical, diagonal high frequency coefficient are processed by the fractional-order calculus mask template that can adaptively select local area differential order, and they are multiplied by the gray-level similarity factor. At last, the enhanced image by wavelet transform is obtained. Experiments shows that this algorithm can achieve a better imagery enhancement effect to some extent.