松辽平原黑土区Aster数据光谱特征及自动分类研究

为研究松辽平原黑土退化状况，以Aster数据光谱特征作为选择最佳波段组合的理论依据，人机交互处理达到计算机自动分类的目的。通过多波段假彩色合成对比发现，Aster数据可见光／近红外波段3N、2、1的组合效果最佳，选择这种组合进行分类有助于提高分类精度。研究区分类结果表明：最大似然分类法总体精度比最小距离法约高6％，两种分类方法配合使用可综合提高分类精度。

Research on spectral characteristics of aster image in black soil area of Songliao Plain and their automatic classification

In order to study the black soil degradation in Songliao Plain, the authors chose the optimal wave band based on the spectral characteristics of Aster image, and realized the computerized automatic classification by means of man-machlne interactive processing. The results show that visible/near-IR band 3N, 2, 1 of Aster image are best for the combination of wave bands through comparing pseudo color composite of multiple bands, which is the best wave band combination in improving the classified accuracy. The consequence of classification shows that the classified effect of maximum likelihood method is much better than minimum distance method; the mutual use of the two methods can greatly improve the classified accuracy.