基于微分搜索的高光谱图像非线性解混算法

发布时间：2018-04-23 18:34

本文选题：高光谱图像 + 谱解混　；参考：《电子学报》2017年02期

【摘要】：针对线性混合模型在实际高光谱图像解混过程中的局限性,提出一种新的基于微分搜索的非线性高光谱图像解混算法.在广义双线性模型的基础上采用重构误差作为解混的目标函数,将非线性解混问题转化为最优化问题.将目标函数中的待求参数映射为微分搜索过程中的位置变量,利用微分搜索算法对目标函数进行优化求解.在求解过程中,通过执行搜索范围控制等机制满足高光谱图像解混的约束要求,进而求得丰度系数和非线性参数,实现非线性高光谱图像解混.仿真数据和真实遥感数据实验结果表明,所提出的非线性解混算法可以有效克服线性模型下解混算法的局限性,避免了由于使用梯度类优化方法而易陷入局部收敛的问题,较之其它高光谱图像解混算法具有更好的解混精度.
[Abstract]:In view of the limitation of the linear hybrid model in the actual hyperspectral image mixing process, a new nonlinear hyperspectral image mixing algorithm based on differential search is proposed. Based on the generalized bilinear model, the reconstruction error is used as the objective function of the mixed solution, and the nonlinear problem is transformed into the optimization problem. The objective function is made by the objective function. The parameters in the process are mapped to the position variables in the differential search process, and the differential search algorithm is used to optimize the target function. In the process, the constraint requirements of the hyperspectral image are satisfied by the execution of the search range control mechanism, and the abundance coefficient and the nonlinear parameters are obtained, and the nonlinear hyperspectral image solution is realized. The experimental results of simulation data and real remote sensing data show that the proposed nonlinear mixing algorithm can effectively overcome the limitation of the solution mixing algorithm under linear model and avoid the problem of local convergence because of the gradient class optimization method, and has a better solution precision than the other hyperspectral image mixing algorithm.

【作者单位】：天津大学精密仪器与光电子工程学院;天津商业大学信息工程学院;光电信息技术教育部重点实验室;河北工业大学电子信息工程学院;
【基金】：国家自然科学基金(No.61401307) 中国博士后科学基金(No.2014M561184) 天津市应用基础与前沿技术研究计划项目(No15JCYBJC17100)
【分类号】：TP751

【相似文献】