求解高光谱解混的三算子分裂方法

打开文本图片集

摘要：  对于高光谱解混问题，提出了一个相应的三算子单调包含和求解该问题的一个分裂方法，其中邻近因子可以自适应选取。该方法还可用于求解更一般的带有线性复合的三算子单调包含问题。数值实验表明，该算法的性能远远超过最近提出的邻近内点方法，并且与其变尺度版本相当。

关键词： 邻近内点法； 单调包含； 分裂方法； 高光谱解混； 自适应

中图分类号： O221.2

文献标志码： A

文章编号： 1671-6841（2025）02-0085-04

DOI：  10.13705/j.issn.1671-6841.2023172

A Three-operator Splitting Method for Solving Hyperspectral Unmixing

DONG  Yunda，  ZHANG  Yuanyuan，  LI  Yiyi

（School of Mathematics and Statistics， Zhengzhou University， Zhengzhou 450001， China）

Abstract： We propose A three-operator monotone inclusion for hyperspectral unmixing was proposed and a corresponding splitting algorithm was given in which proximal factors could be selected adaptively. The algorithm could be also used to solve more general three-operator monotone inclusion with one linear composition. Numerical experiments showed that the algorithm outperformed the recently proposed proximal interior point algorithm and was comparable to its variable metric version.

Key words： proximal interior point algorithm; monotone inclusion; splitting method; hyperspectral unmixing; self-adaption

0 引言

高光谱解混（hyperspectral unmixing）就是将传感器获取的混合像元的光谱分解为不同纯地物的光谱以及估计其相应的比例。（剩余4793字）