首先,针对高空间分辨率卫星遥感影像用于地理信息更新的特点,对信息融合、变化检测、数据挖掘、影像检索等问题进行了研究。遥感影像用于地理信息更新时,一方面应对多源遥感信息进行融合,特别是发展决策层融合算法,另一方面应进行多时相遥感信息融合和变化检测。其次,针对多时相遥感信息融合,提出了一种新的多时相融合处理方法体系,并针对自动信息更新的要求,探讨了遥感数据挖掘应重点解决的问题及其在地理信息自动更新中的应用。最后,面向快速影像提取和检索的需求,研究了基于内容的遥感影像检索技术,并重点分析了检索中能够应用的影像特征。

Oriented to the demands and properties of applying Remote Sensing (RS) image to update geographical information, such topics as information fusion, change detection, data mining and image retrieval were investigated. When RS image was used to update geographical information, both multi-source and multi-temporal RS image fusion were important. In multi-source fusion, decision-level fusion should be emphasized. A new framework of multi-temporal RS information fusion and dynamic analysis including four levels was proposed. According to the demands of automatic information updating, RS Data Mining is proposed, and some key problems were further discussed, also the applications of RS Data Mining to geographical information updating were analyzed. Finally, content-based RS image retrieval used to extract interested images from image database was discussed, and those useful image features were discussed in detail.