本文利用混合像元线性分解方法(LSMM)，对澳门ETM＋图像(2003/1/10)进行像元分解提取植被信息．同时利用同一图像的归一化植被指数(NDVI)、缨帽变换的绿度分量(KT2)对提取的植被信息进行对比分析,发现用LSMM方法提取的植被信息与NDVI的相关系数达到0.93与KT2的相关系数达到了0.74．同时发现用LSMM方法提取的植被面积(4.19 km2)比NDVI阈值法、KT2阈值法提取的植被面积(分别为8.26 km2 8.68 km2)更接近真实植被面积(5.79 km2)．结果表明混合像元线性分解方法能有效地提取植被信息，比以像元为单位的常规遥感提取方法精度更高，为快速、准确、高效的植被监测提供了新思路．
In this paper, the vegetation information of Macao was quantificationally extracted from Landsat ETM+ data of 2003 by using linear spectral un-mixing approach (LSMM). At the same time, the Normalized Difference Vegetation Index (NDVI) image and greenness image (TC2) which was obtained by the Tasseledcap Trasform. These two images, also obtained based on the ETM+ image (2003) of Macao, were used to as two important comparison indexes to evaluate the extracted vegetation
information by LSMM. The result shows that the three images have high correlation. At the same time, the vegetation areas extracted by LSMM, NDVI and TC2 are 4.19 km2, 8.26 km2 and 8.68 km2 respectively. The areas by LSMM are closer the actual vegetable areas (5.79 km2). The results prove the linear spectral un-mixing approach is not only an efficient way to extract vegetation information but also a more accurate measure than routine pixel-based remote sensing methods. LSMM provides a novel way for monitoring vegetation more accurately and efficiently.