节点文献
基于奇异值分解的图像去噪
Image De-noising Based on Singular Value Decomposition
【摘要】 提出了利用奇异值分解去除图像噪声的方法。从矩阵的角度出发,通过对图像矩阵进行奇异值分解,将包含图像信息的矩阵分解到一系列奇异值和奇异值矢量对应的子空间中,然后通过有效奇异值重构图像矩阵达到去噪目的。试验利用MATLAB通过对MRI(核磁共振)医学图像进行去噪处理,验证了奇异值分解的去噪效果,并且通过对多幅图像的试验结果进行分析,得到了去噪重构图像时所需有效奇异值数目的统计值。
【Abstract】 A method for image de-noising based on singular value decomposition is presented. Seeing image as a matrix, the method applied singular value decomposition on image matrix to decompose image information to a series of sub-spaces correspond to singular value and singular value vector, and then by reconstruction of image matrix using effective singular value to remove noising. The results of experiment using MATLAB and MRI image showed that this method could remove image noising and improve signal-to-noise ratio of the image. And also by analysis on experimental results of many images, the statistical value of effective singular value number when reconstructing of image matrix has been gotten.
【Key words】 singular value decomposition; image decomposition; image de-noising;
- 【文献出处】 微电子学与计算机 ,Microelectronics & Computer , 编辑部邮箱 ,2007年11期
- 【分类号】TP391.41
- 【被引频次】20
- 【下载频次】1211