In order to achieve the full range detection of the product structure quality quickly and accurately,first the SIFT algorithm and binary search were selected to find the optimal solution of product which was photographed at any angle in the standard image library.Sec-ondly,according to the prior knowledge turn the product to the angle where the area be identified is and use projection method to segment the region of interest.Finally,images subtract and the calculation of correlation degree is used to determine whether there is defect.The experi-mental result shows that the matching method is used in this paper can save an average of 2.08sthan the traditional fixed-step detection in the premise of ensuring the accuracy of detection.%为了准确快速实现对产品结构质量的全方位完整检测，首先采用 SIFT 算法与折半查找法确定任意角度拍摄的待检产品在标准图像库中最优位置信息，其次根据先验知识将其转到待识别区域所在角度利用投影法分割出感兴趣区域，最后通过减影法及相关度计算判别有无缺陷；实验表明在保证检测准确率的前提下，文章所用匹配方法比传统全周向固定步长方法平均可节省2．08s。
展开▼
