基于Shearlet和稀疏表示的遥感图像融合

针对NSCT变换算法具有较高复杂度、计算时间长、不符合实时性要求的问题,提出将基于Shearlet变换和稀疏表示的算法引入到遥感图像融合中。首先,对待融合图像进行Shearlet变换,分解后得到的低频子带系数采用区域能量取大的融合规则;分解后的高频子带系数采用PCNN的融合规则,最后对重构系数进行Shearlet逆变换。实验结果表明,与NSCT变换及经典算法相比,新方法不仅有效改善了图像融合质量,同时提升了算法的运行速度,实时性良好。

Remote sensing image fusion based on Shearlet and Sparse Representation 1

According to NSCT with high complexity, long computing time andSnon-conformance to real-time, this paper puts forward introducing algorithm based on Shearlet transform and sparse representation into remote sensing image fusion.Through Shearlet transform original images are divided into low and high frequency subbands. Using the rule of regional energy to fuse the former and the rule of PCNN to fuse the latter; Finally, the new multi spectral image can be obtained by the Shearlet inverse transform.Experimental results show that compared with the NSCT and classical algorithm, the new method not only effectively improves the quality of image fusion, but also increases the speed of the algorithm and has good real-time performance.