An FCM-based method to recognize and extract ripe tomato for harvesting robotic system

In harvest robotic system with vision process, extracting ripe fruit from uncertain background is an important issue. In this paper, an FCM (Fuzzy C-Means)-based method combining with mathematical morphology is proposed, while tomato images getting from greenhouse are used to verify the proposed method. The image getting from the vision sensor is color image in our system. Therefore, CIE L*a*b* color space is selected to express the color image in this proposed method at first. Then, the segmentation for the effective color feature component is made using FCM segmentation method. After that, the component with the desired characteristics can be obtained and transferred into binary image for further processing. Lastly, the mathematical morphology method with geometric characteristic is used to extract the largest connected component as the ideal result, and to mark the center and bound rectangle of the recognized fruit. Experiment results indicate that the proposed method achieved good performance.