Skeleton extraction algorithm on grapevine based on machine vision

JIA Ting-meng， XUN Yi， BAO Guan-jun， DONG Mao， YANG Qing-hua
（Key Laboratory of E & M， Ministry of Education & Zhejiang Province， Zhejiang University of Technology，
Hangzhou 310014， China）

Abstract： In order to solve the problem that the bud point position is hard to detect because of the uneven thickness and the irregular shape of the bud，the skeleton extraction method was applied to the vines feature detection. The color image of grapevine was captured indoor. The blue component of the color image was selected for the image preprocessing. Mean filter was used to remove the noise. Threshold segmentation was used to eliminate the wire and shadow and get the binary image of the grapevine. Then morphologic thinning，Zhang thinning，Rosenfeld thinning were used to extract the skeleton of vines binary image. The thinning algorithms were compared to analyze the effect. The results indicate that the skeleton extracted by Rosenfeld thinning gets close to the original shape with well connectivity and centrality. This algorithm lays the foundation of grapevine bud point detection.
Key words： machine vision； image thinning； grapevine； skeleton extraction algorithm； Rosenfeld