Research on Adhesive Workpiece Recognition and Positioning Based on Machine Vision

Abstract

For solving the problem of identification and location of scattered and adhesive workpieces, this paper studies on multi-object workpieces in the background of robot workbench with machine vision technology: Firstly, to segment the adhesive workpieces and extract them, the identification processing is based on the improved SURF(speeded up robust features)+BOW(visual word bag)+SVM(support vector machine). Finally, using depth sensor Kinect to locate the workpiece. Workpiece with different shapes are tested, results show that the system can segment the adhesive workpiece better, and achieve the extraction of multi-target workpiece, while the positioning error is less than 5mm.