Research On Vision Recognition Technique For High-speed Sorting Robot

Machine vision technology has brought profound changes to the traditional industries. As one of the maturity applications with machine vision technology, Industrial sorting reflects this change. Sorting systems which equipped with vision recognition equipment is flexible, efficient and widely used in food and drug production, machining, chip manufacturing and other fields.This topic comes from a project called"high-speed sorting robot", which was generated by Mechanical Engineering College of Jiangnan University. In this paper, based on the needs analysis of the sorting system, we researched on the issues related to the design of vision recognition algorithm and software, as well as the choice of image acquisition device.First of all, for flexibility considerations, this paper presents a formula called "main program and Subprograms" for the experimental system. Then we give an implementation of the program based on the QT cross-platform application framework and the embedded database SQLite.Secondly, based on the needs analysis to the sorting system, and threshold segmentation, edge detection, curve fitting, feature extraction and template matching techniques which provided by Halcon machine vision software package, we designed a series of algorithms for object positioning in different scenes.Finally, we give four examples to verify the positioning algorithm.This paper may have some value on the similar vision system.