Due to the complex operation environment of picking robot,picking efficiency cannot meet the actual production re-quirements.Based on the architecture of picking robot,night recognition and location method based on machine vision is pro-posed.The independent component analysis method based on particle swarm optimization is used to reduce the noise in the night apple image,then the PCNN segmentation method is used to segment the image and the target contour is extracted by edge detec-tion,etc.Finally,the target fruit is improved by the improved three-point circle method Positioning.The feasibility of this meth-od was verified by simulation.The results show that the recognition rate is 94.3%when the occlusion is less than 50%and 89.05%when the occlusion is greater than 50%,which can effectively improve the accuracy of recognition and location.It pro-vides a reference for the development of robot recognition and positioning technology.
关键词
机械视觉/采摘机械人/识别与定位/独立成分分析/三点定圆法
Key words
Mechanical Vision/Picking Robot/Recognition and Positioning/Imperialist Competitive Algorithm/Three Point Circle
维普
万方数据