This work takes the Mecanum wheel mobile robot as experimental platform, Microsoft Visual 2010 and OpenCV (Open Source Computer Vision Library) based on 2.4.9 platform for the development of research on mobile robot vision system. The main work is as follows: first, build the Mecanum wheel robot platform, realizes the mobile robot with full mobility. Second, the Kinect vision sensor is used to acquire the image information in the robot’s visual field and to process the image. Including Canny edge detection, histogram equalization, Harris corner detection algorithm, sub-pixel level corner detection, localization, image affine transformation, Hof line transform, data processing center. Third, the obtained image is processed, and the image quality is improved by image preprocessing, so that the image clarity is improved. Based on the above image information processing, the algorithm is improved on the basis of Harris corner detection algorithm, and the corner coordinate is obtained at sub-pixel precision. Fourth, the processed image information is decided and analyzed in the database of the data center to form the motion control command of the mobile robot. The feasibility of the system is verified by experiments.