Research On Design And Implementation Of Visual Inspection System For Metal Burr Based On FPGA

In the machining of metals industry, burr is a common phenomenon and it affects the accuracy of workpiece size, shape and position accuracy and surface roughness directly. With the further develop ment of machinery manufacturing, both the quality and quantity of the workpieces are critical, especially for the miniature metal workpieces. In the step of removing burrs,there always requires many times removal operation on all machined surface to ensure a very low probability of the existence of burrs. In order to improve the traditional detection methods which relies on artificial method, this paper proposes the use of machine vision technology to detect metal workpiece edge burrs. As a programmable, parallel processing hardware platform, FPGA(Field Programmable Gate Array) has been got more and more attention from researchers in the field of video image processing, and it provides a new processor technology for the visual inspection. To the problem of detection burrs on miniature metal workpiece edge, a depth study has been done based on FPGA core processing platform. The author's major research works are outlined as follows:According to the miniature workpiece edge, it proposed preprocessing algorithm and burrs detection algorithm for the workpiece image. Through analyzing the distribution of the workpiece edge and burr features, a traditional edge detection operators has been applied to achieve the extraction of the workpiece edge, and this paper also used morphological methods to remove noise after the extraction to reduce interference for detection algorithm. By analyzing the binary image of the workpiece's edge, a method based on data-driven edge information statistical anomaly detection algorithms is proposed to achieve the goal of the workpiece edge burr detection and localization.A visual image acquisition and processing systems has been built with the FPGA as the core processor device and the preprocessing and metal burr detection algorithm have been implemented in it with stable performance. A real-time image acquisition and caching module is implemented to facilitate real- time image caching and processing. In this paper, parallel pipeline architecture has been used in the image processing. The communication between each algorithm module was defined by a standard timing interface which reduced the time overhead greatly. Finally, the detection results were displayed on LCD to achieve a real-time online detection for the image of metal parts.Algorithm analysis and test results shows that, at the set lighting conditions, there is a good performance of detection the workpiece edge burr position by FPGA-based vision inspection system. And also the test results show that the burr detection algorithms have very good robustness for different workpiece edge forms. This vision inspection system has been achieved the desired detection purposes.