基于最优识别区间的变步长产品表面缺陷检测研究  

Variable Step Detection of Product Surface Defect Based on Optimal Identification Interval

作　　者：徐青[1] 韩跃平[1] 杨志刚[1] 孙宝华[1] 

机构地区：[1]中北大学信息与通信工程学院,山西太原030051

出　　处：《激光与光电子学进展》2013年第10期86-92,共7页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(61171178;61171179);山西省自然科学基金(2012011010-3);2012年山西省高等学校优秀青年学术带头人支持计划

摘　　要：为实现对产品表面多个待检区域准确快速的检测,在最优识别区间内采用变步长机制快速获取待测产品的周向方位图像序列,实现在有限方位下利用不完全数据对多个待识别区域的快速检测。首先通过相关度计算及投影法确定各待检区域的最优识别区间和旋转步长,其次采用尺度不变特征变换(SIFT)算法与折半查找法确定随机摆放的待检产品在标准库中的最优位置信息,最后通过相关度计算判别各区域有无缺陷。实验表明在保证检测准确率的前提下,基于最优识别区间的变步长方法比传统全周向固定步长检测平均可节省6.37s。In order to achieve the detection of multiple regions in the product surface rapidly and accurately, we select variable step mechanism to obtain the circumferential image sequence of the products to be tested in the optimal recognition interval quickly so that some incomplete data could be utilized to realize the rapid detection of multiple areas in the limited orientation. First of all, we need to determine the optimal identification range of the area and the rotation step through relativity calculation and projection method. Secondly, the scale invariant feature transform （SIFT） algorithm and binary search method are selected to find the optimal location information of product in the standard image library. Finally, we distinguish there are some defects or not by relativity calculation. The experiment results demonstrate that the variable step size method based on optimal identification interval can save an average time of 6.37 s than the traditional full- circumferential fixed step method while ensuring the detection accuracy.

关 键 词：图像处理 缺陷检测 最优识别区间 变步长采样 尺度不变特征变换算法 折半查找法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]