Abstract: In order to meet the requirements of chip recognition and classification statistics, a chip recognition and classification method based on improved Otsu algorithm is proposed to solve the problem of light nonuniformity in illumination system. By judging the global gray value, the threshold segmentation method can be adaptively selected to achieve the purpose of identifying different background color chip boxes. The method of improving the adaptive threshold of Otsu was adopted to eliminate the influence of illumination unevenness. By using the prior information of chip contour, the final extracted chip contour can be modified to ensure the integrity of each chip. Experimental results show that the algorithm can effectively extract the contour of chips in different types of chip boxes, and achieve chip recognition and classification.