Key Technology Research Of Drug Granules Defect Detection System Based On Machine Vision

With the continuous improvement of production technology, the traditional manualdetection for drug tablets had been unable to meet the needs of modern industry.Besides, because the machine vision system can quickly obtain a lot of information,easy to automatically processed, and it is easy to integrate it with design informationand processing information.Thus, machine vision systems are widely used in monitor working conditions,product testing, quality control and other fields in modern automated production process.This thesis describes the research status of machine vision and development trend ofmachine vision inspection technology; then it developed a systematically overall designscheme according to the features of defect detecting for drug tablets; then did a study ofthe system algorithm in depth and made a good attempt on defect detection for medicinecapsules in vc6.0platform, and gained a good result. The main contents of this thesisare as followed:1. It proposed an overall design scheme of the system, and made a detailed analysisand design for the hardware and software structure of the system as well. Besides, itbuilt a drug particle detection platform and gave the system research ideas.2. It conducted in-depth research and analysis on system software algorithms aboutimage processing, such as image enhancement, image segmentation, edge detectionalgorithm and other grains in the drug testing platform on computer terminals, thenchoose the most suitable algorithm contrast and made an appropriate improvements.3. For four different drug tablets, respectively, using a different imagesegmentation algorithm in order to get the best image segmentation.○1Damaged drugparticle detection, it made a good improvement to the edge detection algorithm based onlocal gray level differences, with settling the issue for threshold setting sensitive ofSUSAN edge detection, and using Freeman chain code contour tracking algorithm toquantify Statistics damaged area of drug particles in order to achieve the purpose of removing substandard medicine tablets;○2Shaped tablets drug testing. It can achieve agood detection result of drug tablets profiled with ellipse detection algorithm based onHough transform;○3Drug particle surface defect detection, the use of iterativesegmentation and mathematical morphology can achieve a better screening anddetection of scratches;○4Bubble inside the capsules detection, the application ofmulti-scale adaptive Canny edge detection algorithm can get well bubble contour andthe size of bubble will be quantified.This thesis focuses on four kinds of medicine capsules defect detection and thedetection algorithm and quantification calculations were studied, the research resultshave important guiding significance for the application of machine vision technology todetect defects on drug tablets.