顾及端元光谱可变与空间相关的混合像元扩展模型研究

项目名称： 顾及端元光谱可变与空间相关的混合像元扩展模型研究

项目编号： No.40801181

项目类型： 青年科学基金项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 吴波

作者单位： 福州大学

项目金额： 19万元

中文摘要： 本研究的结果主要包括以下四个方面: (1) 端元光谱可变的混合像元分解。提出利用光谱多尺度分割特征来消除或减弱端元光谱的可变性。(2) 顾及空间关系的混合像元扩展模型及模型。提出一种基于组分空间相关的混合像元分解模型以及混合像元组分的可视化分析与评价方法。(3) 混合像元亚像元目标定位与超分辨制图研究。分别提出利用高斯马尔柯夫模型和多点模拟技术刻画混合像元内亚像元的纹理信息进行亚像元目标定位与超分辨制图;(4)相关应用研究。

中文关键词： 混合像元; 空间相关; 亚像元定位; 端元; 光谱特征

英文摘要： The results og this programme can be summarlized as the following four contents:(1) For the variable end-member spectra of mixture pixels, we have proposed to reduce or mittgate the spectral variabilities by multi scale spactral feature extraction to improve abundance estimation. (2) For spatial dependent mixture model, we have proposed using Gaussian Markov Random Field (GMRF) model to describe fractional abundance spatial denpendence and addressed a unified model based on spectral and spatial information to improve the accuracy of spectral mixture analysis. Moreover, some visual techniques were proposed and used to analyze and evaluate the unmixing accuracies of mixture pixels to help insight spectrally mixture model. (3) For sub-pixel allocation and super- resolution mapping, we have proposed super-resolution mapping methods based on minimization of energy functions defined by spatial dependence either with Gaussian-Markov random field or multi-point simulation technique, to capture the spatially anisotropic structure of sub-pixel land cover classes; (4) some related applications.

英文关键词： mixed pixel; spatial dependence; sub-pixel allocation; end-member; spectral feature