An edge detection algorithm is developed for coal gangue images, and the method has two advantages compared with traditional ones. Firstly, multi-resolution analysis of wavelet transform can improve the quality of edge detection. Secondly, the algorithm is faster for real time. Since the threshold directly from the coefficients of wavelet transform, the rate of recognition for coal gangue is highly raised. The experiment results show that the method is an efficient edge detection algorithm for extraction edges from the noised images of coal gangues.