Research On Defect Recognition Algorithm Of Button With Patterns Based On Machine Vision

Product quality inspection is a very important and indispensable part of the manufacturing process of industrial products. The traditional manual quality control has a lot of disadvantages, such as high labor intensity, low detection efficiency, poor real-time performance and so on. In this paper, based on the DSP smart camera platform, the algorithm have been successfully designed to detect some surface defects, such as edge defects, plugging holes, pits and so on.The paper firstly describes the hardware design scheme of the whole system, then gives a detailed description of the theory and process for three algorithms to detect edge defects, holes and pits respectively. The key of this paper is the research of histogram analysis which is used to do threshold segmentation, thus effectively excluding the interference of irregular patterns, and improving the adaptability and stability of the algorithm; When dealing with color images, characteristic parameters of buttons' contour and surface defects can be extracted. The paper also explores the method to learn characteristics and to make classification standard. Once the algorithm has been designed, it first runs successfully on the VC platform and then is transplanted to smart camera platform. The code must be optimized to greatly improve the detection efficiency. The button samples are tested and the running speed and the detection rate are analyzed.Test results show that the algorithm can effectively identify broken defects, stucked holes defects and pits defects. The average accuracy rate reachs more than 94%, and the average detection rate is 6fps. In a word, the system can accomplish the defects detection task of buttons with patterns.