正交雅可比-傅里叶矩的亚像素边缘定位

崔继文¹, 陈世丽¹, 周彤²

1.哈尔滨工业大学超精密光电仪器工程研究所;2.黑龙江省计量检定测试院



摘要:

为了进一步提高图像的亚像素边缘定位精度,提出了一种基于高阶次正交雅可比-傅里叶矩(OJFM’s)的边缘定位算法.该算法利用OJFM’s的幅值旋转不变性和独特的小图像描述能力,在充分提取图像边缘细节信息的同时抑制图像噪声的影响,对图像边缘旋转后垂直方向上3个不同阶次的OJFM’s之间关系进行分析,将图像轮廓定位至亚像素级.结果表明,该方法定位精度高,对标准直线、标准曲线、CCD采集的实验图像的定位精度分别达到0.05、0.08、0.13像素,可以实现对边缘的高精度定位.

关键词:  图像处理  边缘模型  亚像素定位  高阶正交雅可比-傅里叶矩

DOI：10.11918/j.issn.0367-6234.2011.05.013

分类号:TP391.41

基金项目:国家自然科学基金资助项目(50705026);国家高技术研究发展计划资助项目(2008AA04Z308)



Edge subpixel location based on orthogonal Jacobi-Fourier moments

CUI Ji-wen¹, CHEN Shi-li¹, ZHOU Tong²

1.Institute of Ultra-precision on Optoelectronic Instrument Engineering,Harbin Institute of Technology,150001 Harbin,China;2.Heilongjiang Metrology Institute of Measurement and Verification,150036 Harbin,China

Abstract:

To further improve the edge subpixel location accuracy of an image,a algorithm for edge location based on high-order orthogonal Jacobi-Fourier moments(OJFM’s) is proposed.The rotation invariance and special features of Orthogonal Jacobi-Fourier moments are used to fully extract the characteristics of an edge,while the noises in image is efficiently diminished.The edge location is accomplished by analyzing the relationships among three moments of the edge with setting the edge in vertical direction.The effectiveness of the proposed algorithm is verified through experiment,and the location accuracy is up to 0.05 pixels,0.08 pixels and 0.13 pixels for standard line,standard circle image and actual experimental image respectively.

Key words:  image processing  edge model  subpixel location  high-order OJFM’s