摘要
选取安徽省西南部的沿江平原和皖南山区为研究区,以Landsat8 OLI影像为数据源,基于面向对象的思想,比较阈值法、谱间关系法、NDWI指数、MNDWI指数、EWI指数法、CIWI指数法、PRWI指数法在平原地区和山地地区的水体信息提取的优势与不足。试验结果表明,7种方法均可以有效提取,综合来看,阈值法较适合平原地区的水体信息提取,谱间关系法适合山地地区的水体信息提取,MNDWI指数和阈值结合水体指数法较适合平原地区的水体信息提取。从精度评价角度看,阈值法提取OLI影像的水体信息效果最差,MNDWI指数提取的水体信息精度最高。
This study took plain area and mountain area in Anhui province as an exsample and based on Landsat Oli image,comparatively studied from three aspects: discrimination,water error rate and area accuracy on the single-band threshold method,the multi-band spectral relationship method,NDWI and other four normalized water indexes: MNDWI,EWI,CIWI and PRWI,to show accuracy rating of extracting water information. The results indicated that all of six methods can extact water information. But the single-band threshold method,MNDWI and PRWI are better at extracting water information from plain area. The multi-band spectral relationship method is better at extracting water information from mountain area. On area accuracy asprect,the single-band threshold method is the worst method of extracting water information of OLI image. MNDWI is the best method of extracting water information of OLI image.
出处
《江西科学》
2015年第5期660-665,共6页
Jiangxi Science
基金
江西省自然科学基金项目(20132BAB201049)
测绘地理信息江西省研究生创新教育基地项目