A New Image Processing Enabled Approach for Detection of Scratch Defects for Wire-Type Objects

This paper presents an approach for the surface inspection of wire-type objects. The research aims to detect scratch defect of a wire rod. Scratch defect is caused by many reasons such as poor quality of raw material or malfunction of rolling process. The defects are very difficult to detect accurately due to the scale-covered background and uneven illumination. The inspection system is based on vision technology. Image processing algorithms are applied for detecting scratch defects of wire rod. The detection algorithm is mainly divided into three steps. First, use wavelet transform to have texture analysis. In order to obtain more details, we take the approximation subimage at resolution level one as source image. Then, explain the Fast Line Segment Detector method and the detected results are illustrated using real defects from steel wire rods. But the complex backgrounds in images of wire rods inevitably result in false positive and false negative defects. Finally, apply postprocessing based on detection results. Compared with the classical methods, the results show that the applied algorithms have a good performance on scratch defect detection of wire rod. Moreover, it has an effective execution time proportional to the number of pixels in the image.