MULTIVARIATE ALTERATION DETECTION (MAD) AND MAF POSTPROCESSING IN MULTISPECTRAL, BITEMPORAL IMAGE DATA - NEW APPROACHES TO CHANGE DETECTIONSTUDIES

This article introduces the multivariate alteration detection (MAD) tr ansformation which is based on the established canonical correlations analysis. It also proposes using postprocessing of tile change detecte d by the MAD variate using maximum autocorrelation factor (MAF) analys is. The MAD and the combined MAF/MAD transformations are invariant to linear scaling. Therefore, they are insensititve, for example, differe nces in gain settings in a measuring device, or to linear radiometric and atmospheric correction schemes. Other multivariate change detectio n schemes described are principal component type analyses of simple di fference images. Case studies with AHVRR and Landsat MSS data using si mple linear stretching and masking of the change images show the usefu lness of the new MAD and MAF/MAD change detection schemer. Ground trut h observations confirm the detected changes. A simple simulation of a no-change situation shows the accuracy of the MAD and MAF/MAD transfor mations compared to principal components based methods. (C) Elsevier S cience Inc., 1988.