关键词: 水体信息, 精细化提取, GF-1, 支持向量机, 三峡库区

Abstract: In order to extract the water body information of the long and narrow water area, based on the data of GF-1, and by combining the support vector machine and the visual interpretation, the water body information of the Three Gorges Reservoir and Chongqing City was extracted in this paper. Then, the overall classification accuracy, Kappa coefficient, commission errors, omission errors?, the drawing accuracy and the user’s accuracy were used to validate the results of the rough extraction of water body in the reservoir area. The results showed that the overall classification accuracy of the information extraction in the four experimental areas is more than 90%, except that the Kappa coefficient of the test area 4 is 0.8841, the rest of the test area is more than 0.9, and the extraction accuracy is high. According to the method of visual interpretation on the basis of the rough extraction, the information is refined. The water body information of the Three Gorges Reservoir and Chongqing are obtained with high accuracy and good integrity, and the effective data can provide reference for the fine remote sensing service of this region.

Key words: water body information, refined extraction, GF-1 satellite, support vector machine, high resolution satellite, Three Gorges reservoir area