摘要
从卫星遥感影像中快速、准确地提取水体信息已成为水资源调查,水资源宏观监测及湿地保护的重要手段,但目前,进行水体提取所使用的卫星遥感数据的分辨率较低,使其应用的深度和广度受到严重约束。本文利用波段运算得到特征波段(PRWI),经过图像增强处理后提取水体和居民地混合信息,在此基础上通过分析水体的光谱特征,发现水体和居民地在近红外和短波红外上有显著差异,采用决策树模型将水体专题信息提取出来。运用目视判读和定量统计方法来评价提取结果,结果表明,该方法的总体提取效果较好,其提取精度与常规的监督分类方法相比有了较大的提高。经检验发现其误判的像元主要是位于水体和其他地物的交界处。
Extraction water information from remotely sensed images is signi fi cant for surveying, planning and protecting water resources. In this paper, taki ng Dama county of Tongxiang city as a case study area, SPOT- 5 image collected on Aug, 2002 was used. The approach of extraction is discussed. Firstly, the spe cial band(PRWI) was applied to extract mixed water and residential information. Secondly, the mechanism and characteristics of water bodies and other objects in SPOT- 5 imagery is analyzed to find the possibility of extracting farmland fro m the background. The results show water bodies are distinguishable form residen tial information in SW band and IR band. Thirdly, based on the analysis, a simpl e model of decision tree is applied to extract water information. Finally, the r esults are checked by visual and statistical accuracy assessment. The results su ggest that the model of decision tree is simple and effective and much higher th an the supervised classification's. However, some pixels in the neighborhood bet ween water bodies and mulberry are judged by mistakes.
出处
《上海交通大学学报（农业科学版）》
2005年第2期198-201,共4页
Journal of Shanghai Jiaotong University(Agricultural Science)
基金
国家自然科学基金(40201021)
浙江省科技厅(2004c33089)资助课题
