摘要
水产养殖地已经成为海洋环境监测的热点目标之一。采用具有高光谱分辨率和较高空间分辨率(15m)的ASTER遥感影像,以九龙江河口地区为研究示范区,进行近海水产养殖信息的自动提取方法研究。结果表明,利用ASTER影像的光谱信息和水产养殖地的纹理结构信息,可以实现近海水产养殖地的自动提取。先利用监督分类方法提取混淆有其他水体的水产养殖信息,采用邻域分析来增强水产养殖地的空间纹理信息。通过综合监督分类和水产养殖地空间纹理增强的结果,在专家决策分类器中建立决策规则,进行水产养殖地的自动提取,提取的精度达93%。
It has been one of hot spot targets for extracting aquaculture during monitoring marine environment. In order to study a method of automatic extracting aquaculture land from remote sensing image. ASTER remote sensing image,which is a characteristic of high spatial resolution（ 15 m） and high spectral resolution, was used and the mouth of the Jiulongjiang River was chosen as a case. The results showed that the aquaculture information was able to be acquired automatic when combining spectral and spatial texture configuration information fully, instead of manual interpretation. Steps were as follows. First, aquaculture land mixing other water was extracted by supervised classification; second, spatial texture information was enhanced by neighborhood analysis; finally, decision rules for extracting aquaculture land was built and executed in knowledge engineer classification. There was an accuracy of aquaculture land by 93 percent, which was 13 percent up than supervised classification only, when the method above was applied to study area, it was spatial texture analysis that played an important role during extracting the aquaculture thematic information automatic. A threshold was the easiest to be ascertained by adopting the result from neighbor analysis to the third band of ASTER image. Because it produced the best result based on Expert Classifier, which can integrate general classification and threshold method from the third band in the result from neighbor analysis, we can extract automatically acqufarm information by combining spectrum information and spatial texture enhancement information of auqufarm based on Aster image. However, It may be inappropriate that applies the method to inner aquaculture, which need more effective data and means.
出处
《湿地科学》
CSCD
2006年第1期64-68,共5页
Wetland Science
基金
国家重大基础研究项目(973项目)前期研究专项(2003CCA02100)
福建省科技三项基金(K04016)资助。
