具有训练、学习和零件识别能力的计算机视觉系统

本文提出的系统主要为了在自动化传输带上进行零件的自动识别、定位、定向。曾用该系统对三十多种钟表零件进行反复验证,效果甚佳。

The chief purposes of the system proposed in this pape rare automatic recogni-tion,location,and orientation of machine parts on an automated assembling line.It is provided in this system in accordance with different kinds of machine parts twosets of image feature extraction programs.Establishment of the linked list datastructure for the pattern data effectively facilitates recognizing and training.Thesystem adopts supervisor learning system.It is provided also with programs formanagement and service.All algorithms are written in FORTRAN-Ⅳ-PLUS andtested by more than thirty timepiece parts in the computer vision equipment whichis composed of PDP-11/23 minicomputer,TN2500 CID camera,and TV display.