摘要:An apple recognition method with (R-G) color difference was presented. The improved OTSU algorithm segmented image to increase the operation speed. Through mathematical morphology processing and region filling, repairing and marking, the centroid of connected region was calculated. The experimental results show that under the influence of different light and background, apples that are not shaded seriously can be better identified. The total recognition rate reaches 86%. The color difference information, the improved OTSU algorithm and centroid acquisition provides a visual basis for realizing accurate identification and positioning of fruit and vegetable robot.