基于小波和Zernike矩的标尺靶亚像素检测算法  

Sub-pixel Detection Algorithm for Rule Targets Based on Wavelet Transform and Zernike Moments

作　　者：王会峰[1] 刘永奎[1] 汪大宝[2] 

机构地区：[1]长安大学电子与控制工程学院,西安710064 [2]西安电子科技大学技术物理学院,西安710071

出　　处：《半导体光电》2010年第6期926-930,共5页Semiconductor Optoelectronics

基　　金：国家自然科学基金项目(60872136);长安大学基础研究计划基金项目(CHD2009JC047)

摘　　要：为解决特殊环境下基于光电成像的高精度测量系统中的低照度含噪靶面高精度测量的难题,提出了一种基于小波变换和Zernike矩相结合的亚像素边缘检测方法。该算法首先对获得的低照度含噪靶面图像采用基于小波变换局部模极大值的方法进行抗噪声的粗级边缘提取和去噪处理,获得靶面图像的像素级边缘和具有边缘保持特性的无噪声靶面图像,然后通过在边缘区域内求取Zernike矩的方法进一步提高边缘检测的精度,使得边缘检测的精度达到亚像素级,以便进一步提高测量精度。试验和仿真结果表明:该算法能够在特殊环境下实现对低照度含噪靶面的高精度边缘检测和测量,检测精度达到0.2pixel,具有较强的工程应用价值。In order to make high accuracy edge detection of noise-contained images that are brought on by special image environment, new sub-pixel edge detection algorithm is introduced based on wavelet transform and Zernike Moments. In the algorithm, the low lumen and noise-contained target images are firstly edge extracted based on wavelet transform modular maximum, which can resist noise infection to edge detection and a pixel-degree edge of the images can be detected. Then the Zernike Moments are divided in the edge areas which can improve the edge detection precision to a sub-pixel degree. Tests and simulations show this algorithm can get high precision edge detections in low lumen noise-included environment and the detection precision can reach 0. 2 pixel.

关 键 词：成像测量 边缘检测 亚像素 小波变换模极大值 噪声 ZERNIKE矩 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]