基于块匹配和共同向量的图像去噪研究

Image Denoising Study Based on Block Matching and Common Vector

针对图像去噪过程中存在边缘细节信息丢失的问题,提出一种结合块匹配和共同向量的图像去噪方法。通过块匹配在像素块邻域内搜寻相似块构建向量集,由Gram-Schmidt求出共同向量,再由原始向量与共同向量的差经过线性最小均方误差估计后,与共同向量重建图像块。实验结果表明,方法能够有效去除图像的高斯白噪声,并能保持图像的边缘细节信息。

To solve the problem of missing the edge detail information in the image denoising process,a new image denoising method based on block matching and common vector is proposed. First,it searched for similar blocks to construct the vector set in pixel block neighborhood based on the block matching algorithm. Second,the common vector was calculated by Gram-Schmidt algorithm. Then the differential vectors were obtained from the original vector and common vector and calculated the linear minimum mean square error estimation. Finally,with the common vector the image blocks was reconstructed. Through many experiments,the results show that method can effectively remove the gaussian noise image,and can keep the image edge details.