中文说明：
在工业生产中总是经常遇到裂痕、划痕和变色等产品的表面缺陷问题，而这些问题不管对于人工检测还是机器视觉检测都极富挑战。其难度在于该类缺陷形状不规则、深浅对比度低，而且往往会被产品表面的自然纹理或图案所干扰。因此，表面缺陷检测对于正确打光、相机分辨率、被检测部件与工业相机的相对位置、复杂的机器视觉算法等要求非常高。 机器视觉划痕检测的基本分析过程分为两步：首先，确定检测产品表面是否有划痕，其次，在确定被分析图像上存在划痕之后，对划痕进行提取。
English Description：
Surface defects such as cracks, scratches and discoloration are often encountered in industrial production, and these problems are challenging for both manual inspection and machine vision inspection. The difficulty lies in the irregular shape of the defect, low depth contrast, and is often disturbed by the natural texture or pattern on the surface of the product. Therefore, the requirements of surface defect detection are very high, such as correct lighting, camera resolution, relative position of detected parts and industrial cameras, and complex machine vision algorithm. The basic analysis process of scratch detection in machine vision is divided into two steps: firstly, to determine whether there is scratch on the surface of the detected product, and secondly, to extract the scratch after confirming the existence of scratch on the analyzed image.