Research on the Key Technologies of Vision Detection in Workpiece Flaw
|School||Hefei University of Technology|
|Course||Signal and Information Processing|
|Keywords||Defect detection Machine Vision Image processing DSP|
Visual inspection system has been widely applied to the workpiece defects detection has the advantages of speed , high precision , non-contact , can effectively overcome the lack of artificial detection , increase the degree of automation of the workpiece production and production efficiency . The visual inspection of defects of the workpiece is the the workpiece detection intelligent key link . Control bottles defect detection . The main content of this paper and the innovation lies : the image acquisition system performance analysis and classification summarize experimental platform to build the image acquisition , the workpiece image of the real-time , high-quality collection . 2 based on the image geometry regularization , NSCT (Nonsubsampled Contourlet Transform) for image processing . Based on NSCT image multi-directional multi-resolution decomposition , combined with semi - soft threshold denoising , the signal-to-noise ratio significantly improve ; combined modulus maxima image edge detection , the method to extract the edge is smooth and complete with strong noise resistance . 3 the the workpiece defect detection methods and identification of the type of defect . The defect detection the Hough transform positioning the center of the circle detection method based on template matching method mouthpiece ; through the defect perimeter , area, and extending the length and the degree of circularity of the four parameters measured to identify the bottle defect types . Through a large number of experiments to verify the feasibility of DSP-based image processing algorithms developed image acquisition and processing procedures , the system software porting from the PC to the DSP . Dynamic real-time detection of consecutive images to explore and study and design a user - friendly operator interface .