基于Bayesian压缩感知的融合算法

摘要

Acorrding to the sampling pattern in compressive sensing theory, this paper proposed an image fusion method in the compressive domain based on an improved sampling pattern. Firstly, it used double-star sampling pattern to get the compressed measurements in the sparsity domain for image fusion, then it used a simple and efficient maximum absolute value rule as the fusion rule, at last, it used the Bayesian analysis method to reconstruct the fusion image. Simulation results verify the effect of the proposed method.%根据压缩感知理论中的采样模式,提出了一种基于改进采样模式的压缩域图像融合算法.该算法首先通过双星型采样模式获得待融合图像的稀疏域压缩测量值,然后利用一种简单的绝对值最大融合规则直接在压缩感知域进行融合,最后通过贝叶斯的方法重构融合图像.在图像重建的过程中采用了贝叶斯方法.由于考虑了误差以及噪声的影响,使得融合图像的质量进一步提高.仿真结果表明,该算法具有良好的融合效果.