Image Processing Algorithm for Developing Selective Harvester of Cabbage

Abstract

This paper describes a high speed color image processing algorithm which has been developed to apply to a robotic cabbage harvester. The developed algorithm is used to recognize the cabbage heads by processing the color image which is taken under unstable lighting conditions in the field. The algorithm uses Neural Network to extract the cabbage heads of HSI transferred image, and estimates the location and the size by calculating the correlation between a binalized image and two templates of model cabbage. Image processing speed is improved by optimizing the algorithm and using parallel processor. As a result of experiment, it could judge 70% of cabbage heads for selective harvest, and it took 8.8s to recognize 11 cabbage heads.