-
摘要: 针对北方农牧交错地区地表异质性和破碎化强烈、地物光谱特征混杂严重、土地利用/覆被遥感信息提取对目视解译依赖程度高的特点,以内蒙古多伦县为研究区,尝试将基于知识的遥感信息提取技术应用于该地区。通过地物光谱特征的深入分析,用线性光谱混合分解模型将主要地物覆被类型分离,并建立多个专题信息模型,依据经验知识建立了各用地类型提取规则,对多伦县土地利用信息进行了计算机自动提取,并与单一的监督分类法提取的结果进行了对比。结果表明,运用遥感信息提取技术提取的面积精度达93.8%,抽样点评价精度为87.7%;单纯使用监督分类法提取的面积精度为71.1%,抽样点评价精度为59.5%。可见,遥感信息提取技术较单一的监督分类法效果好很多,实现了多伦县土地利用遥感信息高精度分层次快速提取。图4表3参20Abstract: The heterogeneity and fragmentation of land were serious and the spectrum was complicated in northern farming-pastoral regions,so the land use and cover classification research was highly dependent on visual interpretation. Duolun County was selected as a test area and the knowledge-based remote sensing information-extracting technology was applied to this area. The main cover types were decomposed by linear spectral mixture model. And a number of thematic information models were set up by the analysis of spectrum. The extraction rules of all cover types were set up based on empirical knowledge,and then the land use information of Duolun was extracted automatically on a computer. The results showed that the area precision by this method was 93.8% and the sampling point precision was 87.7%. The corresponding results of supervised classification were 71.1% and 59.5%t. It was proved that the method was more effective than supervised classification and could realize the fast extraction of remote sensing information with high precision and multiple tiers.[Ch,4 fig. 3 tab. 20 ref.]
-
Key words:
- land resource management /
- remote sensing /
- expert knowledge /
- spectral mixture analysis /
- Duolun County
-
链接本文:
https://zlxb.zafu.edu.cn/article/doi/10.11833/j.issn.2095-0756.2010.03.016
计量
- 文章访问数: 4238
- HTML全文浏览量: 275
- PDF下载量: 223
- 被引次数: 0