The Research Of PET Bottles Packaging Defect Detection Based On Image Matching Algorithm

With social progress and the improvement of living standards, people on the quality of consumer goods also made increasing demands, the quality of products has become a manufacturer and consumer concerns. In recent years, huge market demand for the mineral water, fruit juice drinks, such as the production of a wide variety of industries maintained steady and rapid development. Beverage bottles and packaging quality will have a direct or indirect impact on the quality of its internal beverages; machine vision applications of PET bottles detection system can greatly reduce testing costs, improve product quality, and speed up the production speed and efficiency. Therefore, the low-cost, high precision and high efficiency of the PET bottles detection system to study a very important practical significance.In this paper, the research applies to the industrial pipeline on the PET bottle packaging quality inspection system. According to PET bottle packaging and quality testing of the characteristics of the actual production requirements, the first objective analysis of the design of the system's overall programme design, including the overall structure of the system design, system hardware choices and the conduct of its software system the design and implementation.In the image matching algorithm, in SUSAN edge detection algorithm on the basis of their improvements, a new adaptive to extract its poor gray threshold of improved SUSAN edge detection algorithm, to achieve a better edge detection results. PET bottles detection system in accordance with the characteristics, from space and features similar to the perspective of the measure, proposed the use of images of gray, in the form of information and transform domain of information, through comprehensive structural characteristics of images to match, because SUSAN detection algorithm Characteristics of accurate positioning, and not sensitive to local noise, and the image histogram moments with the same translation, rotation, the proportion of invariance, so this form of image-based information, information and moments of gray histogram image matching the same moment to improve Algorithm. In ensuring the accuracy of the match at the same time, greatly enhance the speed match, both have strong anti-gray, anti-geometric distortion capacity, but also has strong noise suppression.Finally, the PET bottles and caps quality testing for the realization of quality inspection and testing in accordance with the actual situation on the detection algorithm for the corresponding improvements, the experimental results show that the algorithm can achieve real-time online testing requirements.