论文部分内容阅读
传统的图像融合算法,如IHS变换、合成变量比、主成分替换等方法所得到的融合图像通常存在不同程度的光谱扭曲现象.一种基于混合像元分解的图像融合算法(FSMA)可以很好地保持图像的光谱信息,但该算法仅在以终端端元为先验知识的模拟数据中得到了成功的应用.分析表明,由于全色波段与多光谱波段光谱响应函数不同以及多光谱波段通常不能覆盖整个全色范围,原有的FSMA算法并不能直接应用到真实遥感数据中.文中提出了一种改进的基于混合像元分解的图像融合算法(IFSMA).该算法通过重构原有算法中优化问题的目标函数,降低了对利用多光谱数据模拟全色波段亮度值的难度,使得基于混合像元分解的图像融合算法可以推广应用到真实遥感数据中.实验结果表明,IFSMA算法在光谱信息和空间信息的保持方面均优于IHS变换、合成变量比、主成分替换以及原有的FSMA等算法.
The traditional image fusion algorithms, such as IHS transform, synthetic variable ratio, principal component substitution and other methods, often result in spectral distortions in the fused images.A FSM algorithm based on mixed pixel decomposition can be very good But the algorithm is only successfully applied to the simulation data which is prior knowledge of the terminal endmember.The analysis shows that due to the difference of the spectral response function between the panchromatic band and the multispectral band and the spectral response of the multispectral band Can not cover the whole panchromatic range, the original FSMA algorithm can not be directly applied to the real remote sensing data.An improved fusion algorithm based on mixed pixel decomposition (IFSMA) is proposed in this paper.Through the reconstruction of the original algorithm The objective function of the optimization problem reduces the difficulty of simulating the luminance value of the panchromatic band by using the multispectral data so that the image fusion algorithm based on the mixed pixel decomposition can be widely applied to the real remote sensing data.The experimental results show that the IFSMA algorithm has the advantages of good performance in the spectrum Information and spatial information are better than the IHS transform, synthetic variable ratio, the replacement of the principal component and the original FSM A and other algorithms.