新的噪声污染灰度图像边缘检测方法

New method for edge detection of noisy gray scale image

传统的图像边缘检测方法由于引入了各种微分运算,因此用于噪声图像边缘检测时对噪声极度敏感。针对这一问题,提出了一种基于独立分量分析技术的噪声图像边缘检测方法,该算法通过计算数据之间的高阶统计信息,提取特征模板,然后将被高斯噪声污染的灰度图像与这些模板逐个匹配,提取出边缘成分。实验结果表明,基于独立分量分析技术的模板匹配方法自适应强,复杂度低,是一种有效的高斯噪声污染灰度图像边缘检测方法。

The normal methods with different algorithm are very sensitive to noise in edge detection.According to this problem,a new edge detection method for noisy gray scale image based on Independent Component Analysis （ICA） is proposed.The methods extract pattern template by calculating the higher statistical information of image data.Then a gray scale image contaminated by Gaussian noise is transformed into a pattern map by pattern matching.Thus the edges can be easily detected from pattern map.The experiments show that the proposed algorithm is a valid method of edge detection for image contaminated by Gaussian Noise.