【摘要】 由于中低空间分辨率的遥感数据在一个像元中记录了两种以上地类的不同光谱 ,从而获得了两个以上地类的光谱之和以及混合像元 ,当地类破碎时混合像元越多 ,而且在两类以上地类的交界处混合像元最多 .该文模拟不同空间分辨率的遥感图像中混合像元造成的误分类情况 ,同时应用线性模型对混合像元进行了分解提纯 ,用以提高分辨率精度更多还原

【Abstract】 In middle or low resolution remote sensing image, a pixel is consisted of more than two land use classes, so its spectral value is the sum of these land use spectrum, and it is called mixed pixel. The more fragmentation of land classes, the more mixed pixels, there are more mixed pixels on the boundary of different classes. This article simulates misclassification because of mixed pixels on different spatial resolution images, then applies linear model to decompose mixed pixel, so as to improve classification accuracy.更多还原