In order to apply the image processing technique to the surface defects online-detecting of stamping parts, a real-time fast detection system is designed. Multi-pattern matching algorithm is used to locate the stamping parts in the image, then the region of interest is built. The shading correction algorithm based on the Laplace-Gaussian (LoG) operator is proposed to enhance the defect parts of workpieces. The Otsu algorithm and morphology are used to extract the defect parts. The system adopts MATLAB to realize the filtering algorithm based on LoG operator, the rest of the algorithm are implemented by LabVIEW and the MATLAB script node is called by LabVIEW, the multithreading technology is used to accelerate the computation. According to the experiment results, the proposed system can detect each stamping parts on the production line and detect the defectives. The whole process takes less than 100 ms, which can satisfy the demand of online detection.