Vision-based detection of tomato main stem in greenhouse with red rope

In order to identify tomato plants for target spraying, an algorithm was presented to detect main stem of tomato relative to the rope which was used to fix main stem. The distribution characteristics of tomato images due to HSI color space were analyzed, and the images were then binarized using Otsu segmentation method based on H histogram and the rope region was extracted. The rope line was fit with least square method based on the set of discrete points extracted by thinning methodologies. Experiment results indicated that the average processing time for each image of 640×480 pixels was 0.16 s, the recognition accuracy of 100 images was 93%, and the maximum deviation between the rope and tomato main stem was 48 pixels. The algorithm can detect the main stem accurately with strong robust.