| Abstract [eng] |
This paper describes the design and research of a plastic bottle caps defect inspecting machine vision system. The purpose of this vision system is to detect and analyze defects in plastic bottle caps, eliminating the use of human resources, that is, to automate the process of quality assurance. Thus, the object of the project research is a plastic bottle caps defect inspecting machine vision system. In this project, plastic bottle caps of seven different colors were analyzed, in which a defect may occur at the edge or gasket. The aim of the project is to design and research of a plastic bottle caps defect inspecting machine vision system. Project objectives: to analyze the methods of image recognition used in industry, the methods of calibration of industrial video cameras and the influence of different types of lighting for image recognition; select the appropriate equipment for image analysis; design a lighting system; create a plastic bottle caps defect analysis program; to study the repeatability of the analysis results and the factors influencing analysis time duration; to statistically investigate the effectiveness of plastic caps defect detection. Research methods: analysis of scientific literature, experimental research. In the analysis of the scientific literature, the similarities and differences, advantages and disadvantages of machine vision and human vision were reviewed first. Industrial cameras, their calibration methods, lighting systems, and image processing and analysis tools were also reviewed to select the correct vision system components. The design part describes the entire structure of the vision system, consisting of selected components (area scan (2D) camera with VGA resolution, 6 mm focal length lens, Intel NUC mini-computer for analysis, etc.). Lighting system design (modeling of light incidence angle, creation of luminaire layout and lighting level simulations with DIALux software) is described in detail. Finally, the algorithm of the image analysis program (developed using the Matrox Design Assistant software) and the problems encountered during the development of the program (e.g. incorrect color detection) are presented. The research part describes the calculation of the repeatability of the analyzed data (97.38%), the time of each step of the program algorithm and the total execution time (35.86–49.69 ms) for all seven plastic cap colors. Finally, twenty different cases of defect inspection of plastic bottle caps are described in detail, in four cases the defect is difficult to detect and in one case it is not detectable at all. To correct this case, program modifications were made, which extended (+21.15 ms) the previously measured execution time of the algorithm. |
