NSST域内基于压缩感知和PCNN的遥感图像融合

针对NSCT融合算法运算数据量大、计算复杂度较高、实时性较低的问题,提出一种基于非下采样剪切波变换(nonsubsampled shearlet transform,NSST)和压缩感知(compressed sensing,CS)的遥感图像融合方法。首先,对待融合图像进行NSST变换,分解后得到的低频子带系数采用区域能量的融合规则;分解后的高频子带系数具有较高稀疏性。通过CS进行压缩后采用PCNN的融合规则,最后对重构系数进行NSST逆变换。实验结果表明,与传统经典算法相比,新方法不仅有效提高了图像融合效果,而且加快了算法的运行速度,满足融合系统实时性的要求。

Remote sensing image fusion based on compressed sensing and pcnn in nsst domain 1

According to NSCT with larger operation data , higher computational complexity, lower real-time performance, this paper puts forward a new fusion method based on based on Compressed Sensing and PCNN in NSST domain. The low frequency sub-band coefficients are fused by the rule of regional energy. High frequency subband coefficients of the decomposed with high sparse, compressed by CS then fused with the rule of CS and PCNN.Finally, the new multi spectral image can be obtained by the NSST inverse transform. Experimental results show that compared with the traditional classic algorithms, new method not only effectively enhances the effect of fused image, but accelerates the speed of the algorithm and satisfys the requirement of real time fusion system.