Aiming at the problem involving corners clustering and the difficulty of threshold selection in Harris operator, by analyzing the theory of Harris operator, Harris corner detection algorithm based on self-adapting non-maximal suppression is proposed. Firstly, the algorithm detects the pixels which corner response function is local maximum, then local maximum Harris corners are descendant ranked and a suppression radius for each corner is set up, corners are extracted through the suppression radius continuously decreasing, corner clustering is effectively avoided and spatial even distributed over the image for the Harris corners is realized. At the same time, the algorithm is able to solve the difficult problem of threshold selection and improve the adaptability of algorithm. Experimental results show that the algorithm is able to detect corner more reasonable, and could be used for image mosaic and motion estimation, etc. quite well.