摘要
针对噪声对图像边缘检测等过程带来的极大干扰严重地影响了图像质量的问题,研究图像中所混有的白噪声对边缘定位精度的影响。采用一维灰度矩法与高斯拟合法这两种亚像素边缘定位法进行比较、分析,分别讨论了两种算法在偏移量不变、不同信噪比情况下,噪声对边缘定位精度的影响。实验结果表明,在对受噪声影响的图像进行边缘检测时,采用高斯拟合法能够有效地抑制噪声,提高检测精度,在信噪比SNR≥30dB时的分辨率精度能达到0.02个像素,而一维灰度矩亚像素边缘法无法达到该精度。
Noise interference with image impacts on image quality severely.It brings great difficult for the course of image edge detection.So an important problem during image process is to research precise of edge orientation affected by noise.In the paper,methods of sub-pixel edge orientation such as 1-dimention gray moment algorithm and Gauss fitting method are used.The paper discussed affection of edge orientation precise by noise under two conditions respectively.One is that signal to noise ratio（SNR） and displacement are different using above there methods.The other is that the same SNR using the same method.The research in this paper is greatly valuable in order to restrain noise availably,protect edge details and improve detection precise.When SNR is above or equal 30 dB,resolution precision is 0.02 pixels obtained by using Gauss fitting method rather than 1-dimention gray moment algorithm.
出处
《武汉理工大学学报》
CAS
CSCD
北大核心
2012年第7期141-145,共5页
Journal of Wuhan University of Technology
基金
吉林省教育厅科研项目(2010290)
东北电力大学研究生教改项目(201201)
河南省科技攻关项目(122102310475)
关键词
边缘检测
定位精度
噪声
信噪比
高斯拟合法
edge detection
orientation precision
noise
signal to noise ratio（SNR）
Gauss fitting method
