遥感影像居民地信息自动提取方法研究进展  

RESEARCH PROGRESS OF RESIDENTIAL AREA INFORMATION AUTOMATIC EXTRACTION METHODS BASED ON REMOTE SENSING IMAGE

作　　者：李德一[1] 王大鹏[1] 张安定[1] 王周龙[1] 

机构地区：[1]鲁东大学地理与资源管理学院,山东烟台264025

出　　处：《云南地理环境研究》2006年第5期48-51,共4页Yunnan Geographic Environment Research

基　　金：鲁东大学校基金课题

摘　　要：快速掌握居民地的时空分布和变化特征对于区域可持续发展具有重要意义。遥感技术的发展,为居民地的快速提取提供了新的途径。传统的人工目视解译费时费力,基于遥感影像的居民地信息自动提取势在必行。主要探讨了居民地信息自动提取的5种方法,即基于统计的分类法、谱间结构阀值法、归一化指数法、纹理分析法、地学专家知识分类法,讨论了各种方法的优缺点,总结了目前的研究进展。最后提出了居民地信息自动提取方法的发展趋势,认为多种分类方法相结合,建立居民地信息提取的专家系统将是今后研究的重点。Mastering spatial-temporal distribution and transformation characteristics of residential area in a short time contributed to region sustainable development significantly. The development of remote sensing technology provided a new approach for rapid extraction of residential area. Because traditional manual visual interpretation was time and labor consuming, it was necessary to extract the information based on remote sensing image. This article mainly discussed five methods, such as statistics-based classification method, threshold of spectrum structure method, normalized difference index method, texture analysis method and geo-based expert knowledge classification method. The advantages and disadvantages of each method were discussed; current research progress was also summarized respectively. Finally, some develop trends of residential area information automatic extraction methods were put forward, combined several methods and set up an expert system of residential area information extraction will be the research emphases in the future.

关 键 词：信息提取 归一化建筑指数 纹理 地学专家知识 

分 类 号：X87[环境科学与工程—环境工程]