基于LOG_Zernike算子的快速图像测量.pdf

第 卷第 期 光电工程 Vol.34, No.3 34 3 2007 年 3 月 Opto-Electronic Engineering March, 2007 文章编号：1003-501X(2007)03-0063-05 基于 LOG-Zernike 算子的快速图像测量 葛 继 （江西理工大学 应用科学学院，江西 赣州 341000 ） 摘要：图像测量技术，一个很重要的方面就是图像的边缘检测定位，而边缘定位的准确与否又直接影响到测量的 精度。常用的检测算法都是像素级的，对测量要求较高的场合无法满足要求。定位速度一直是边缘检测算法的一 大瓶颈，本文正是在保证一定精度的前提下，研究了快速图像测量算法，即运用 LOG 算子先对工件进行快速边 缘定位，在此基础上采用 Zernike 矩法将精度提高到亚像素级。为了得到实际的测量尺寸，应用块规进行了系统 标定并进行了一系列测量实验，实验结果表明该方法定位精度接近于 Zernike 矩法，且算法的运行时间大约在 1.5s 左右。 关键词：LOG 算子；Zernike 矩；标定；亚像素定位；图像测量 中图分类号：TP391.4 文献标识码：A Fast image measurement based on LOG-zernike operator GE Ji ( Applied Science School , JiangXi University of Science and Technology, Ganzhou 341000, China ) Abstract ：One of the most important aspects in image measurement technology is edge detection and orientation. Most common detection algorithms based on pixel level can not meet higher measuring requirements. This article studied the fast image measuring algorithm and precision as the premise. Edge was oriented using LOG operator firstly, and then the measuring accuracy was improved to the subpixel level by Zernike moment based on pixel level. To obtain the actual measuring dimension, the system was calibrated with gauge block and a series of measuring experiments were carried out. The results show that orientation accuracy with this method is close to Zernike moments. The runtime of this algorithm is about 1.5s. Key words：LOG operator; Zernike moment; Calibration; Subpixel orientation; Image measurement 引 言 边缘是图像最基本的特征，它包含了大量用于识别的有用信息并为人们描述或识别目标和解释图像提 20 80 供