Capacitor is the necessary component of electronic whole products. Because of manufacturing technology and the limit of equipment's level, the thief is a common capacitor cosmetic defects. This paper puts forward a regional segmentation algorithm of the thief capacitor based on the k-means clustering algorithm and combining the characteristics of the image of the thief capacitor. According to the research results of Ohta, et al, the first component of color feature is chosen and used as the image segmentation by employing the K-means clustering method. This paper defines the number of classification is two, and gives the algorithm of initial clustering center and spacing threshold according to rough set theory. Experimental results show that this image segmentation algorithm can effectively segment the thief area of the capacitor image, and have good precision and accuracy.