Design of Machine Vision Applications in Detection of Defects in High-Speed Bar Copper

In the process of producing copper bar, because of the cast copper billet, rolling equipment, rolling process and other reasons, the surface of copper bar appear some defects such as crack, scarring, roller printing, scratches, holes, scales, pitting, and so on. These deficiencies not only affect the appearance of the product, but more importantly reduce the product's corrosion resistance, abrasion resistance, fatigue limit and other capabilities. We design defect detection system base on machine vision. The detection system uses linear CCD camera to get images, uses LabVIEW in the "NI Vision Development Module" image processing module for image analysis and processing, include image pre-processing (e.g., image segmentation, etc.), binarization, determining the detection area and other treatment methods, get the binarization images of defects on copper surface, extract various characteristic parameters of the binary image, do image recognition, then determine whether flawed. Use "BP neural network" classifiers to achieve non-linear complexity of computing and high-speed parallel processing, and accurately distinguish between the main copper strip appears in defect types. The system can be designed for real-time detecting high-speed movement of the copper-line, it has high processing speed and high precision, it can be extended to other more industries to use.