Adaptive filtering approach to multispectral image processing

Adaptive filtering approach to reduce noise and enhance details readability of Landsat images is presented in the paper. As a method of the filter weights adjusting the Recursive Least Squares algorithm was used. It allows correcting images from different channels according to the reference image. This image is a simple average of images acquired by Landsat 7 ETM+ scanner but more sophisticated reference images can be used. In the analysis the misfit images were uses instead of filtered one. In the paper example of adaptive filtering of images from Cairo City, Egypt is presented. We compare the adaptive filtering results with unsupervised land classification.