A measure is proposed based on the information theory and geostatistics to evaluate information content in remotely sensed images .The method is based on the additive noise model and maximum mutual information T.hese factors affecting the information content have been taken into account ,such as noise , spatial correlation and so on .It is suitable for measuring the information content in optical images that have robust spati al correl ation with different l and cover types .An experiment was performed on a Landsat TM image with three different kinds of l and cover types (city ,farml and and mountain) .The result shows that city has the most information content .It also proves that there is a log positive correl ation between information content and the vari ance of the images .%提出了一种结合信息论与地学统计法的遥感影像信息量计算的方法。此方法基于遥感影像加性噪声模型和互信息上界的计算原理，同时考虑了噪声和空间相关性等影响遥感影像信息量的因素，适于计算具有稳健空间相关性、不同地物类型的光学影像的信息含量。利用Landsat TM影像子集，分别计算了城市、农田、山地3种不同地物类型的影像信息量。结果表明，城市含有最大的信息量，同时影像信息量与影像方差呈对数正相关关系。
展开▼
