针对植被覆盖区的岩性识别，本文在线性混合象元分解的基础上通过对TM影象中混和象元进行补偿置换，即用象元中非植被地物光谱信息置换植被部分，裸露象元在此过程中没有变化，高植被象元（包括全植被象元）被掩膜，从而达到定量消除植被影响的目的，同时混合象元中非植被信息得到增强。

In this paper, the compensated replacement method is used to process the mixed pixels in TM image on basis of linear mixing model in order to identify rocks in the area with vegatation cover. The spectral information of vegetation in one mixed pixel is replaced with that of non vegetation in the process. The pixels without vegetation information are not changed and the pixels with heavy vegetation information are masked. The effect of vegetation cover is quantitatively removed and non vegetation information in mixed pixels is strengthened with use of this method.