Ripe Tomato Recognition and Localization for a Tomato Harvesting Robotic System

Abstract

A ripe tomato recognition and localization system for tomato harvesting robotic systems in greenhouse is developed. The ripe tomato is segmented by K-means clustering using the L*a*b* color space. To extract a single ripe tomato, mathematical morphology is used to denoise and handle the situations of tomato overlapping and sheltering. Tomato's shape features are combined with the color features to recognize ripe tomatoes. The difference value between the centroid coordinate and the center coordinate of image is used to control the robot arm to aim the tomato center. The turned angles of the robot arm are recorded. The distance between the tomato and robot arm is measured by a laser sensor. With the turned angles and the distance, the tomato's 3D coordinate is calculated under the spherical coordinate system. Experimental results show the effectiveness of the proposed method.