基于Zernike正交矩的图像亚像素边缘检测算法改进.doc.docVIP

第 XX 卷 第 X 期 自 动 化 学 报 Vol. XX, No. X 200X 年 X 月 ACTA AUTOMATICA SINICA Month, 200X  基于 Zernike 正交矩的图像亚像素边缘检测算法改进 高世一 1; 2 赵明扬 2 张 雷 1; 2 邹媛媛 1; 2 摘 要 介绍了 Zernike 矩及基于 Zernike 矩的图像亚像素边缘检测原理, 针对 Ghosal 提出的基于 Zernike 矩的亚像素图像边缘检测算法检测出的图像存在边缘较粗及边缘亚像素定位精度低等不足, 提出了一种改进算法. 推导了七乘七 Zernike 矩模板系数, 提出一种新的边缘判断依据. 改进的算法能较好检测图像边缘并实现了较高的边缘定位. 最后, 设计 3 组不同试验. 试验结果同 Canny 算子及 Ghosal 算法相比证明了改进算法的优越性. 关键词 Zernike 矩, 亚像素, 边缘检测, 图像处理 中图分类号 TP751 Improved Algorithm about Subpixel Edge Detection of Image based on Zernike Orthogonal Moments GAO Shi-Yi1; 2 ZHAO Ming-Yang2 ZHANG Lei1; 2 ZOU Yuan-Yuan1; 2 Abstract The principle of Zernike moments and the method of sub-pixel edge detection based on Zernike moments were in-troduced in this paper. With the consideration of the limitation of the subpixel edge detection algorithm by Ghosal, such as the lower location precision of the edge and the extracted wider edge than that of the original image, an improved algorithm was pro-posed. On the one hand, a mask of size seven multiply seven was calculated and could be applied for the edge detection. On the other hand, a new criterion for edge detection was put for-ward. Additionally, a series of experiments were designed and implemented. The experiment results show that accuracy of the improved algorithm is higher than that obtained from using Canny algorithm and Ghosal algorithm. Key words Zernike moments, Subpixel, Edge detection, Im-age processing 缘检测的实质是通过一些算法来提取图像中灰度 连续的边缘像素. 常用的边缘检测算子有 Roberts 算 子、Prewitt 算子、Laplacian 算子以及 Canny 算子等.Robert 算子直接计算图像差分检测边缘, 算法简练但不能有效抑制噪声;Prewitt 算子通过取像素平均值对图像进行差分和滤波, 适用于处理灰度渐变和噪声较多的图像, 其缺点是可能 虑掉有用的边缘信息;Laplacian 算子用于检测屋顶型边缘效果较好, 但对噪声比较敏感, 且检测精度较低[1].Canny 算子是基于高斯函数提出的. Canny 算子提取的边缘最为完整,而且边缘的连续性较好, 这主要是因为它进行了非极大值抑 收稿日期 2007-07-27 收修改稿日期 2007-12-04 Received July 27, 2007; in revised form December 04, 2007 国家高技术研究发展 (863) 计划资助 (2006AA04Z235) Supported by the national High-Tech Research and Development Plan of P. R. China (2006AA04Z235) 1. 中国科学院沈阳自动化研究所 沈阳 1