基于稀疏和冗余表示的图像融合方法

An Image Fusion Method Based on Sparse and Redundant Representation

为了产生能提供更多信息的融合图像,提出了一种基于SRR(Sparse and Redundant Represent,稀疏和冗余表示)理论的图像融合方法。该方法首先采用训练所得的冗余字典对原始图像进行稀疏表示,然后对系数矩阵采用1-范数取大融合规则进行融合,最后对融合系数矩阵和冗余字典进行重建以得到融合图像。将该方法与拉普拉斯变换、PCA(Principal Component Analysis,主成分分析)、DWT(Discrete Wavelet Transform,离散小波变换)、CVT(CurVclct Transform,曲波变换)和NSCT(Non-Subsampled Contourlct Transform,非下采样轮廓波变换)等方法在几组图像上进行比较,实验结果表明,该方法在一定程度上提高了融合图像的质量,在主观和客观上都具有较好的性能。

A new image fusion method based on sparse and redundant representation theory is proposed to produce the fusion image which can provide more information.First,the source image is represented with sparse coefficients using an over-complete dictionary.Second,the coefficients arc combined with the choose-max fusion rule.Finally,the fused image is reconstructed from the combined sparse coefficients and the dictionary.The proposed method was compared with methods based on Laplace(LAP),Principal Component Analysis(PCA),Discrete Wavelet Transform(DWT),CurVelet Transform(CVT)and Non-Subsampling Contourlct Transform(NSCT)on several pairs of multi-focus images.The experimental results demonstrated that the proposed approach performed better in both subjective and objective qualities.