Abstract: This paper takes the recognition and location method in the process of watermelon picking is taken as the research object, and the infrared spectrum technology is used to obtain the image and related location information in the process of watermelon picking. Through the way of experimental measurement, the best band of infrared light which can effectively identify watermelon is determined, and the infrared image of watermelon is collected under different lighting conditions, the image is filtered and de-noised, and the infrared image of watermelon is condensed by using the stereotyped template to realize the recognition of watermelon fruit, and the centroid coordinates in the process of watermelon picking are further calculated as the reference watermelon picking point. According to the appearance shape of watermelon, the image is segmented and located, and the cutting points of watermelon are identified. The results show that the success rate of the recognition system is 98%, the success rate of picking point is 93.8%, and the success rate of cutting point is 87.7%.