The Research On Detection And Classification System Of Porcelain Based On Machine Vision

As a technique to achieve automation and modernization of production,computer vision technology has attracted a lot of attention from researchers at homeand abroad arising from the date of birth. It has wide applications in industrial control,biomedical, agricultural engineering, military and other fields, and it is also a popularresearch topic in the field of artificial intelligence. For the machine vision-baseddetection is fast, accurate, non-contact and a series of advantages, it attracts drawnincreasing attention as a modern means of detection.Detection and classification of porcelain based on machine vision has a veryimportant significance to improve the competitiveness of China’s market. Currently,the traditional way to detect the presence of artificial porcelain classification processinefficiencies, high rate of false positives, slow and other shortcomings. It has beendifficult to meet the needs of production, Also it is difficult to meet the requirementsof real-time classification. To solve this problem, to achieve automatic detectionporcelain, to shorten the time the entire workflow, the paper made the followingstudies:1, This paper introduces the working principle-based machine vision system forautomatic detection of porcelain, the overall structure and workflow. Analyze thehardware characteristics of the system, select a suitable light source, graphics card,industrial cameras and other hardware to build a hardware platform detection system.2, Collecting porcelain image, and make smoothing, segmentation, gray andenhanced image processing after image acquisition, meanwhile, a compare severalimage preprocessing methods, including image noise removal method, grayingmethod, image segmentation, edge extraction methods.3, Extract the size, shape, color characteristic parameters from the image.Classify porcelain according to the characteristic parameters. Porcelain size featureextraction using the projected area of investment and improved law; shape featureextraction, using the histogram discrimination law; color feature extraction methodusing primary colors.4, Use BP neural network to be the classifier, after the experimental data wasanalyzed, the results showed that detection and classification system based on machine vision achieve the desired goals.Through the study of computer vision porcelain automatic classification methodbased on automatic classification of objects for the future of the foundation, but theclassification of complex objects is a complicated systematic project, involving anumber of areas and disciplines, because of my time and energy restrictions Thereforethe work done in this paper is still very limited, the prototype tool development is stillin the exploration stage and need further study and research in the future work.