Shearlet与改进PCNN相结合的图像融合

为提高多聚焦和医学图像融合的性能,提出了一种基于Shearlet变换的新型图像融合算法。与小波变换类似,Shearlet具有简单的数学结构,这使其可以很方便地和多分辨分析关联起来。在对一幅图像作Shearlet变换时,可以将其在任意尺度和方向上进行解构,因而Shearlet比传统小波可以捕获更多的方向和其他几何信息。所以对于图像融合来说,Shearlet是一种很好选择。对于Shearlet子带系数的选择,采用了一种改进的PCNN的点火幅度来得到融合策略,而不是传统PCNN方法中的点火次数,点火幅度通过一个Sigmoid函数来得到。并且采用改进拉普拉斯能量和(SML)这一有效的聚焦度量作为PCNN的输入,以提高其性能。实验结果表明,该方法在视觉效果和客观评价指标上都要优于小波和非下采样Contourle(tNSCT)方法。

Image fusion based on Shearlet and improved PCNN

A novel approach is proposed to improve the performance of multifocus and medical image fusion, which is based on Shearlet transform for image fusion. Shearlets are equipped with a simple mathematical structure similar to wavelets, which are associated to a multi-resolution analysis. Moreover an image could be decomposed by Shearlet trans- form in any scale and any direction, in which Shearlets show a greater ability to fully capture directional and other geome- trical features than traditional wavelets, so they are a suitable tool for image fusion. The subband coefficients of Shearlets are selected by fusion decision based on an improved Pulse Coupled Neural Network（PCNN）, which is completed by the PCNN fire amplitude, instead of the total pulse number. While the PCNN fire amplitude is calculated by a sigmoid function. Moreover, SML （Sum Modified Laplacian）, as an effective focus measures, is used for the input of Sigmoid PCNN to improve its performance. Experimental results show that the proposed algorithm outperforms similar competitive strategies based on wavelet and Nonsubsampled Contourlet Transform（NSCT） in terms of both visual quality and objective evaluation.