Contourlet相关性和PCA的图像融合算法

Contourlet变换克服了小波变换在处理高维信号时的不足,它比小波变换具有更好的方向性和更好的稀疏表达性能。将Contourlet变换应用于图像融合领域,能更好地提取图像边缘特征,为融合提取更多的特征信息。基于Contourlet变换系数相关性的图像融合算法是将图像进行Contourlet变换分解后,针对高频分解系数尺度内以及尺度间像素点具有的相关性设计图像融合规则,低频信息选择PCA的方法进行融合,最后通过重构得到融合图像。实验结果表明Contourlet能够为融合图像保留更全面的原始图像信息,基于相关性的图像融合算法能够更加有效、准确地提取图像中的特征,是一种有效可行的图像融合算法。

Image fusion algorithm based on correlation among contourlet coefficients and PCA

Contourlet transform overcomes the weakness of wavelet in higher dimensions. According to the theory of Contourlet, Contourlet can represent the characteristics of image. When Contourlet is applied to image fusion, the characteristic of original images can be effectively extracted and more important information is preserved. The Contourlet coefficients are recognized as independent in traditional image fusion method based on Contourlet. However,the coefficients of Contourlet have strongly dependency among different region and different direction subbands,and using the characteristic of Contourlet coefficients can design fusion rule. Experimental results have evidenced the effectiveness of the proposed method and it can preserve and extract the characteristic more reliable, accuracy and effective.