Design And Implementation Of Detection System For Surface Of Plastic Cups

Aiming at the disadvantages of manual inspection, like slow speed, low accuracyand high cost of human labor, on the surface quality of plastic milk cup of theproduction line in the current domestic. Studies have found that quality detection systembased on the machine vision inspection technology had high speed, high precision,strong anti-interference ability and perfect continuity, etc. So this paper presented theapplication of machine vision to complete the non-contact detection which focusing onstains and labels on the surface of milk cup. The main work and contributions are listedas follows:(1) Combined with plastic milk cup production enterprise’s actual detectionrequirements, this paper gave some detailed analysis of key and difficult point in qualitydetection of the cup, and designed the hardware and software for the quality inspectionsystem from the perspective of machine vision. Among these architectures the main partof hardware for machine vision system were analyzed in detail, such as the IPC,industrial camera, optical lens, light, etc.(2) In order to meet the needs of quality detection, the quality-inspection task ofmilk cup were divided into the outer wall detection and the inner wall detection, whichused an industrial camera separately. Relative to the simplicity of gray distribution ofthe outer wall image, the inner wall image showed more difference at different regions.Therefore, so as to improve the quickness and accuracy of detection algorithm, theregional identification and sub-regional stain detection method was put forward. Thispaper used contour tracking algorithm based on Freeman chain code, regionalidentification algorithm based on vector marking, regional zoom algorithm and circlefitting algorithm respectively to divide the inner wall of the milk cup into5differentregions, the mouth, the edge, the side wall, the bottom edge and the bottom,to be morespecific.(3)According to the result of the division, the corresponding stain detectionalgorithm was presented respectively for characteristics of image in different parts of the inner wall. In view of the changes of gray level in the central regions of the lateralwall and the bottom was smooth and without clear boundaries, so in these areas, staindetection algorithm based on the Sobel filter was used; But the shadows with obviousvariance really existed at the grooves of the edge and bottom and showed clearboundaries. So in this paper, morphological operation and the seed region growthalgorithm were used to segment the suspicious besmirch and then identify the stainsaccording to the regional characteristics. In addition, the detection of the outer wall andthe labels also used the Sobel filter algorithm. And then judging whether stains existedand the labels were qualified or not separately via the features of stains on the outer walland the labels.(4)This paper has developed the software interface for the quality inspectiondetection system based on VC++6.0. The software interface is friendly and easy tooperate. After testing the system, the results have showed that it can meet the user’srequirements and ensured the quality of plastic milk cup on the high-speed productionline.