基于空间加权协同稀疏的高光谱解混算法研究

Spatially weighted collaborative sparse unmixing for hyperspectral images

针对传统稀疏解混算法因空间信息利用不足带来的丰度图像空间分布连续性差的问题,本文提出了一种基于空间加权协同稀疏的解混方法.该方法利用协同稀疏正则项刻画丰度系数的行稀疏性;同时,在协同稀疏框架下,引入空间加权因子挖掘高光谱图像邻域像元间的空间相关性.本模型采用交替方向乘子法求解,通过交替迭代,对空间权重和丰度系数进行优化.模拟和真实高光谱数据实验结果表明本文方法能够比现有同类方法获得更精确的解混结果.

In this paper,we propose a spatially weighted collaborative sparse unmixing method aiming at fully exploiting the spatial information in the hyperspectral images,in which a collaborative sparse regularizer is used to describe the row sparsity of the abundance,while on the top of the collaborative regularizer,a spatial weighting factor introducing the spatial correlations is incorporated. The proposed model is optimized by the well known alternating direction method of multiplier.Our experimental results,conducted using both simulated and real hyperspectral data sets,illustrate the good potential of the proposed algorithm which can greatly improve the abundance estimation results when compared with other advanced sparse unmixing methods.