基于Arimoto熵和Zernike矩的刀具图像亚像素边缘检测  

Sub-Pixel Edge Detection of Cutting Tool Images Based on Arimoto Entropy and Zernike Moment

作　　者：吴一全[1,2] 龙云淋 周杨[1] WU Yi-quan;LONG Yun-lin;ZHOU Yang(College of Electronic and Information Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 211106, Jiangsu,China;Provincial Key Laboratory of Manufacturing and Automation,Xihua University,Chengdu 610039,Sichuan,China)

机构地区：[1]南京航空航天大学电子信息工程学院,江苏南京211106 [2]西华大学制造与自动化省高校重点实验室,四川成都610039

出　　处：《华南理工大学学报（自然科学版）》2017年第12期50-56,共7页Journal of South China University of Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61573183);西华大学制造与自动化省高校重点实验室开放课题(S2jj2014-028)~~

摘　　要：为满足基于机器视觉的刀具尺寸测量系统快速及高精度的要求,提出一种基于直线截距直方图的Arimoto熵和Zernike矩的图像亚像素边缘检测方法.首先,通过高斯滑动窗口获取图像的邻域平均灰度,构造图像的灰度-邻域平均灰度二维直方图,并利用直线截距法将其降为一维直方图;然后,针对得到的直线截距直方图,依据Arimoto熵准则进行阈值分割,并将所得阈值映射回原二维直方图实现目标区域及像素级边缘的提取;最后,由基于Zernike矩的边缘模型对获取的像素级边缘进行重定位,以完成刀具图像亚像素级边缘的提取.通过对刀具图像进行的大量实验,将文中方法与基于Canny的、基于空间矩的、基于灰度矩的以及基于Zernike矩的边缘提取方法进行对比,发现文中方法运行速度更快且提取精度更高.In order to meet the high-speed and high-accuracy demands of the machine vision-based measurement system of cutting tool sizes,an image sub-pixel edge detection method based on Arimoto entropy of linear intercept histograms and Zernike moment is proposed.In the method,first,the neighborhood-average grayscale of images is obtained through the Gaussian sliding window to construct a two-dimensional histogram,and the linear intercept method is adopted to reduce the two-dimensional histogram to a one-dimensional histogram.Then,aiming at the achieved linear intercept histogram,the thresholding is performed according to the Arimoto entropy,and the obtained threshold is mapped back to the two-dimensional histogram to extract a target region and pixel-level edges.Finally,the edge points are re-located by using the Zernike moment-based edge model,thus achieving the sub-pixel-level edges of cutting tool images.By a large number of experiments on the cutting tool images,the proposed method is compared with the Canny-based,the space moment-based,the gray moment-based and the Zernike moment-based edge extraction methods.The results show that the proposed method is superior in both speed and accuracy.

关 键 词：刀具图像 亚像素边缘检测 直线截距直方图 Arimoto熵 ZEMIKE矩 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]