Pvc Surface Defect Detection Based On Image Visual Methods And Applications

PVC product is making up of polyvinyl chloride material, with the constant development of science and technology, surface quality of industrial products is more and more required for the follow-up industry. The inspection and control of surface defects has attracted PVC product manufacturers' close attention. The study is significant practical significance and has wide application prospect in Automated Visual Inspection System which is suitable for PVC product.In this work, on the basis of studying the category of surface defects of PVC product, a hardware system for Automated Visual Inspection and inspection techniques for defects detection was designed. The System comprises image collection (CCD camera, image grabber card, LED surface light) and image understanding (computer) and communication. To the detection technique, according to main defects of geometrical deformation, color distortion, surface contamination and crack and based on the study of image processing technology, techniques used in every procedure of edge inspection and mathematical morphology operation were designed and tested, for better image processing of PVC surface defects and feature extraction. To the key algorithms, Fuzzy Theory is used to the image segmentation techniques, the image segmentation techniques used are based on fuzzy threshold segmentation. To the assignment of a membership degree to traditional fuzzy threshold segmentation is not certain and the window must be selected by manual work, thresholding using type II fuzzy sets and image segmentation based on fuzzy threshold of adaptive window were applied to extract features. To the work of mathematical morphology, the binary image extracted was processed with mathematical morphology operation which was combined to carry out defect label with the initialization of appropriate structure elements.The surface image of PVC product have been shoot on vidicon, Fuzzy Theory is used to the image segmentation techniques, and mathematical morphology operation is used to the image recognition, and the result showed the technology could effectively detect usual surface defects of alveoli or crack or spot on PVC plastics, exhibiting good real-time and robustness. The system could be applied to every on line surface defect inspection for PVC product, to realize real time detection, alarm and control and improve degree of automation and in formalization level.