论文部分内容阅读
由于干旱区农牧交错地带湿地面积小且地类混杂,混合像元现象严重,使得该区湿地信息的自动提取难度较大。针对湿地遥感信息提取的特点和难点,采用NDVI(normalized different vegetation index)阈值法提取水体,应用支持向量机模型(SVM)提取去除水体后的湿地信息,并以修正线性光谱混合分解模型分解的草甸分量,进一步提取高盖度、中盖度和低盖度草甸信息。试验验证结果表明:提取结果的总体精度为88%,Kappa系数为0.83。该方法可为其他光谱特征混杂地区湿地遥感信息的提取提供参考。
Due to the small wetland area and mixed land cover types in the arid area, the phenomenon of mixed pixel is serious, making it difficult to extract the wetland information automatically. Aiming at the characteristics and difficulties of remote sensing information extraction in wetland, the water body was extracted by normalized different vegetation index (NDVI) and the wetland information was extracted by support vector machine model (SVM) Austin component, to further extract high-cover, medium cover and low cover meadow information. The experimental results show that the overall accuracy of the extraction results is 88% and the Kappa coefficient is 0.83. This method can provide reference for the extraction of remote sensing information of other wetlands with mixed spectral features.