Extraction of Litchi Stem Based on Computer Vision under Natural Scene

The aim of the litchi stem extraction based on computer vision is to provide the basic image for the position of litchi plucking point. In this paper, the litchi stem extraction was studied by comparing various color models and combining with the cluster segmentation algorithm under natural scene. For the complex natural scene, three steps were taken to finish the litchi stem extraction. Firstly, the image of string-like litchi fruits was extracted through the C_r component in YC_bC_r color model; Secondly, the image of litchi fruits was extracted from the color image by K-means clustering method which based on RGB; finally, the image of litchi stem was obtained by the image subtraction operation and morphological processing. The results showed that the success rate of stem extraction is 80% by experimenting thirty images collected under natural scene.