基于卷积稀疏表示的图像融合方法

图像融合是图像处理领域中比较重要的一门技术，传统的图像融合方法会降低图像融合质量。针对稀疏表示在图像融合中存在一定的缺陷，提出了一种基于卷积稀疏表示的图像融合方法。首先，对高频子带系数进行合理有效处理，利用相似度分析和视觉显著性进行融合。然后，将低频子带系数整体融合改进为使用Butworth低通滤波对低频子带进行分解，得到低频近似子带和强边缘子带。最后，再用改进的脉冲耦合神经网络(Pulse Coupled Neural Network，PCNN)对强边缘子带进行融合。实验结果表明，与其它传统的图像融合方法相比，信息熵(Information Entropy，IE)提高了将近3%，标准差(Standard Deviation，SD)提高了将近9%，空间频率(Space Frequency，SF)提升了将近30%，互信息(Mutual Information，MI)提升了将近25%。同时，时间效率也有了一定程度地提升。

Image Fusion Method Based on Convolution Sparse Representation

Image fusion is an important technology in the field of image processing, traditional image fusion methods can reduce the quality of image fusion. Aiming at the shortcomings of traditional sparse representation in image fusion, an image fusion method based on convolution sparse representation is proposed. Firstly, the high-frequency sub-band coefficient is reasonably and effectively processed, and the similarity analysis and visual significance are combined. Then, the overall fusion of the low-frequency sub-band coefficient is improved to decompose the low-frequency sub-band by using the Butworth low-pass filter, and the low-frequency approximate sub-band and the strong edge sub-band are obtained. Finally, an improved pulse coupled neural network(PCNN) is used to fuse the strong edge sub-band. The experiment results show that information entropy(IE) has increased by nearly 3%, standard deviation(SD) by nearly 9%, spatial frequency(SF) by nearly 30% and mutual information(MI) by nearly 25% compared with other traditional image fusion methods. At the same time, time efficiency has been improved to a certain extent.