摘要
传统的仅依赖于光谱信息的遥感影像信息提取方法难以满足精度要求,基于CART算法构建规则集,利用决策树技术对玉树地区高分辨率影像进行建筑物信息提取,除了光谱信息外还结合建筑物的纹理特征信息。结果表明,提取精度较传统的最大似然法有了较明显的提高,说明此方法对于高分辨率影像建筑物提取具有一定的可行性。
Traditional information extraction method from remote sensing images depends on spectral information, and is difficult to satisfy accuracy requirement. We combine spectral information with texture feature information to extract building information from high resolution images in Yushu area based on CART algorithm. Results show that comparing with traditional maximum likelihood method, the extraction accuracy has been improved. The method for high resolution image building extraction is feasible.
出处
《地震》
CSCD
北大核心
2013年第2期96-102,共7页
Earthquake
基金
国家高技术研究发展计划(863计划)(2012AA121304)
高分光学卫星遥感地震应急技术研究(E0205/1112/9)
