Contourlet变换与粒子群优化相耦合的遥感图像融合方法

Remote Sensing Images Fusion Method Coupling Contourlet Transform with Particle Swarm Optimization

为有效优化融合图像中多光谱特性的保持和空间信息的保留,提出一种结合Contourlet变换与粒子群优化算法的遥感图像融合方法。通过设定粒子群优化算法的目标适应度函数,使其依赖于融合结果图像的客观评价指标,并自适应地获取低频子带的最优加权系数和高频子带间结构相似度阈值的最优值,从而得到优化的融合图像。首先将全色图像和多光谱图像的亮度I分量分别进行Contourlet变换,根据分解后的低频系数和高频系数不同的特征信息,在低频系数上以信息熵与相对偏差的差值作为目标适应度函数,采用优化算法自适应地寻找最优加权系数进行融合;在高频系数上以结构相似度作为目标适应度函数,搜索结构相似度的最优阈值p,再采用基于区域结构相似度的融合规则进行融合;最后经Contourlet和IHS逆变换得到融合图像。仿真实验结果表明:提出的方法能很好地兼顾多光谱图像光谱信息的保持和全色图像空间信息的保留。

In order to effectively optimize the retention of multi-spectral characteristic and the reservation of spatial information in the fusion image,we presented a remote sensing image fusion method by coupling contourlet transform and particle swarm optimization(PSO).The optimized fusion image is achieved by setting the PSO fitness functions which depend on the objective evaluation indexes of fusion image.The optimal weighting coefficient of lowpass sub-band and the best threshold for the regional structure similarity of highpass sub-band are adaptively obtained by PSO.Firstly,the panchromatic image and I component of multispectral image are decomposed respectively by using the contourlet transform.According to the different feature information,the difference between entropy and relative deviation is regarded as PSO fitness function to adaptively find the best weighting coefficient by the optimized algorithm in the lowpass subband.Meanwhile,the structure similarity is considered as PSO fitness function to search the best threshold pand the fusion rules of regional structure similarity is used to fuse image in the highpass sub-band.Finally,the fused image is reconstructed by inverse transform of contourlet and IHS.The simulation experiment results show that the proposed algorithm can effectively preserve the spectral information and spatial information of the original images.