FAN模型的双种群差分进化光谱解混算法

发布时间：2018-01-20 10:24

本文关键词： 高光谱遥感光谱解混双线性混合模型差分进化算法双种群机制　出处：《遥感学报》2017年02期 　论文类型：期刊论文

【摘要】：双线性混合模型是近年来非线性光谱解混的研究重点之一,其克服了线性混合模型无法描述地物多重散射作用的缺陷,能够更精确地还原真实的地物光谱混合过程。然而,限于模型的复杂性,目前在缺乏准确的端元先验知识的条件下进行双线性光谱解混仍是一项具有挑战性的任务。差分进化算法(DE)是一种具有良好全局搜索能力的群智能优化算法,其优化求解过程无需进行复杂的数学推导,为双线性光谱解混问题提供了一种有效的解决途径。为此,本文以FAN双线性混合模型为例,提出了一种双种群机制的差分进化算法(记为DEFAN),实现非监督双线性光谱解混。DE-FAN算法通过建立端元与丰度两个种群的交替进化机制寻找最优解,同时在迭代中引入自适应重构策略增强种群多样性,降低算法陷入局部最优解的风险,最终实现端元与丰度的同时估计。通过模拟图像及真实图像的解混实验进行算法检验,证明DE-FAN算法较之传统非线性解混算法具有更高的解混精度及解混效率。
[Abstract]:Bilinear mixing model is one of the key points in the research of nonlinear spectral demultiplexing in recent years. It overcomes the defect that linear mixed model can not describe the multi-scattering action of ground objects. The spectral mixing process of real objects can be reduced more accurately. However, it is limited to the complexity of the model. Bilinear spectral unmixing is still a challenging task in the absence of accurate end-element prior knowledge. It is a swarm intelligence optimization algorithm with good global search ability. The optimization process does not need complicated mathematical derivation, which provides an effective way to solve the bilinear spectral unmixing problem. Therefore, the FAN bilinear mixed model is taken as an example in this paper. A differential evolutionary algorithm (DEFAN) based on double population mechanism is proposed. An unsupervised bilinear spectral unmixing. DE-FAN algorithm is implemented to find the optimal solution by establishing an alternative evolutionary mechanism between endmember and abundance. At the same time, adaptive reconstruction strategy is introduced to enhance population diversity. Reduce the risk of the algorithm falling into the local optimal solution, and finally realize the end element and abundance estimation at the same time. Through the simulated image and real image unmixing experiment to verify the algorithm. It is proved that the DE-FAN algorithm has higher precision and efficiency than the traditional nonlinear unmixing algorithm.
【作者单位】：华南师范大学地理科学学院;广东水利电力职业技术学院计算机信息工程系;
【基金】：国家自然科学基金(编号:61673184,41571349,40901232) 高分辨率对地观测系统重大专项~~
【分类号】：TP751;TP18
【正文快照】： 引用格式:张锐豪,罗文斐,钟亮,覃事银,李倩倩.2017.FAN模型的双种群差分进化光谱解混算法.遥感学报,21(2):239 252Zhang R H,Luo W F,Zhong L,Qin S Y and Li Q Q.2017.Double-population differential evolution algorithm forbilinear spectral unmixing based on FAN mode

【相似文献】