<正>数据压缩处理是利用电子计算机,对遥感图象进行模式识别的一项重要技术。本文应用Karhunen—Love变换与R型因子分析,对陆地卫星计算机兼容磁带的多光谱数据作了压缩处理。结果,在基本保持遥感信息量的情况下,使特征量由四维压缩成二维,因而就能在座标纸上作出点聚图,对事物加以识别与区分。从计算结果看来,R型因子分析比Karhunen—Loève变换似更理想。

With the aid of electronic computers, data compression processing is an important technique used for recognizing the pattern of remote sensing images. In this paper, Karhunen-Loeve transformation and the R-Factor Analysis have been used in compression-processing the multi-spectrum data on the land satellite Computer Compatible Tape (CCT). As a result, with about 99% of the information being kept, the characteristic quantities can be compressed from four dimensions into two dimensions, so that a picture with clusters of dots can be made on a sheet of paper for recognizing and defining the objects.