本文选题来源于国家高技术研究发展计划（2007AA12Z156），教育部新世纪优秀人才支持计划（NCET-06-0131）和国家自然科学基金（N0.40672195）。遥感数据作为一种重要的空间数据源，已经在环境监测、资源管理、灾害预报、重大工程监理、国防安全等领域发挥着不可或缺的作用。目前，多种统计方法（主成分分析、因子分析、相关性分析、回归分析等）和模式识别、人工智能都在空间信息挖掘中取得了一定的效果。然而这些方法仅考虑了数据的低阶统计特征，对数据源的限制较多，并不能满足空间信息的复杂性。稀疏成分分析(Sparse Component Analysis, SCA)是将一系列随机变量表示成稀疏线性组合的方法，相关理论进入快速发展阶段，目前尚难以见到任何应用分析方面的文献。因此，我们在空间信息挖掘中引入了一种新的在信号处理领域也是热点的盲源分离方法—稀疏成分分析。论文取得的主要成果如下：1.讨论了基于稀疏度量的稀疏成分分析，重点研究了过完备的稀疏成分分析、自适应稀疏成分分析等几种稀疏成分分析算法，并以广东省肇庆市遥感数据(TM)为例，探讨了基于稀疏成分分析在遥感图像分类中的应用，实验结果表明，基于稀疏成分分析的遥感图像分类结果明显优于主成份分析，总体分类精度提高了14.87%。2．探讨了遗传算法与稀疏成分分析结合的问题。

The paper is supported by National high-tech research development plan (2007AA12Z156), New-Century Training Programme Foundation for the Talents by the State Education Commission（NCET-06-0131）, the National Natural Science Foundation (NO.40672195).Remote sensing data as an important source of spatial data, which has been in environmental monitoring, resource management, disaster forecast, a major project supervision, national security,is playing an indispensable role. Many statistic methods including Principal Component Analysis, Factor Analysis etc. have got some effect in geological information mining. However, these methods just consider the low order statistical feathers of data, and have many limitations of the data sources, cannot satisfy with the complexity of spatial geological information.Sparse component analysis is a method of a series of random variables into a sparse linear combination. Theory into the stage is rapid, yet it is difficult to see any application to the analysis of the literature. We introduce into a novel blind sources separation method—Sparse Component Analysis.The research work in this thesis is as follows: 1. This article discusses the sparse measurement in sparse component analysis, focused on the over-complete and adaptive sparse component analysis algorithm. We take Zhaoqing City, Guangdong Province of remote sensing data (TM) as an example of sparse component analysis based image classification in remote sensing application, experimental results show that the remote sensing image classification results based on sparse component analysis are better than principal component analysis, the overall classification accuracy increased 14.87%. 2. In addition, the paper also discusses the genetic algorithm combined with the sparse component analysis problem.