改进的流形学习图像稀疏降噪方法

Image Denoising Based on Improved Sparse Representation with Manifold Learning

本文将稀疏重构与流形学习算法两算法结合运用于图像降噪方面,提出了基于拉普拉斯图谱嵌入的稀疏编码。该方法利用拉普拉斯图谱的局部相关性,通过对权重矩阵的改进,增强数据间的关系表示,同时又通过稀疏理论进一步优化代表低维数据点的稀疏系数进行数据压缩,从而进一步提高图像降噪效果。

Integrating sparse representation with manifold learning,a novel algorithm for sparse coding based on Laplacian eigenmap embedding was proposed.By use of the local correlation of Laplacian structure with the improved weight matrix,this method can represent the relationship of each data point more effectively.On the other hand,by using sparse theory,the method can further optimize the sparse coefficients which are present as the low-dimension data points.The results demonstrate that the proposed method can achieve a better performance in image denoising.