Identification And Location Of Multi-class Hybrid Wire Harness Terminals Based On Machine Vision

As far as industrial detection technology is concerned,machine vision detection technology is an inevitable trend in its future development,but its application in the wire harness terminal industry is still quite lagging,especially for the identification and capture of wire harness terminals.Most of the traditional technical schemes complete the arrangement and positioning of the wire harness terminals through the mechanical way of the vibrating disc which is not conducive to the flexible production of the enterprise.In view of this situation,this paper adopts machine vision technology to guide robot,one-time global positioning and grab coordinate guidance,so as to improve flexibility and enterprise flexible production capacity.The main work and research results are as follows:(1)Image acquisition system and data set construction.According to the actual needs of the enterprise,the image acquisition system is constructed including the lighting device,the industrial camera which is calibrated and the lens selection design.Under the condition that the calibration precision meets the requirements of the project,the image data of the wire harness terminal is constructed to prepare for the subsequent algorithm design.(2)Image segmentation of harness terminals.The states presented by the harness terminals are mainly the states of abutment,superposition,isolation,steady,mctastable,nesting and so on.The enterprise guarantees that there is no such state through mechanical devices for the problems of superposition state,steady state,metastable state and abutment state.A segmentation method based on gray threshold is adopted for isolated wire harness terminals.An OTSU threshold segmentation algorithm with mask function is designed for single class attached wire harness terminals.A multi-class matching wire harness terminal is designed for a template matching algorithm based on contour features.(3)Image classification of harness terminals Identification.In this paper,two methods are proposed to realize the classification and recognition of isolated wire harness terminals:fusion of geometric and shape features and directional gradient histogram.For single-type wire harness terminals,a template matching algorithm based on local features is used to realize classification recognition.For multi-type wire harness terminals,a template matching method based on contour features is adopted to segment the contour of each independent wiring harness terminal,then it is classified and identified according to the classification and recognition method of isolated wiring harness terminals.Finally,the accuracy and real time of the algorithm are verified(4)Image positioning of harness terminals.An adaptive detection algorithm for locating the datum line is designed.According to the positioning datum line,the location feature point of the wire harness terminal is intercepted to calculate the location center point and the manipulator grab rotation angle of the wire harness terminal.The comprehensive positioning accuracy can reach 99.36%through offline testing(5)Software system development.According to the actual production demand of the enterprise,the classification and positioning algorithm of the fusion geometric feature and the shape feature are selected to develop the software system by considering the classification accuracy and the efficiency of the algorithm synthetically.The system mainly integrates image acquisition system,data preservation module and communication module.wire harness terminal positioning accuracy of 99.025%meets the enterprise requirements for positioning accuracy(greater than 99%)through the online test.