Change Detection from Remote Sensing Imageries Using Spectral Change Vector Analysis

Abstract

Change detection using remote sensing data is the process of identifying and examining temporal, spatial and spectral changes of pixel signal. This paper detected land cover changes from two Landsat ETM+ imageries using spectral change vector analysis (CVA). CVA is a change detection technique that can determine the direction and magnitude of changes in multidimensional spectral vector. In this paper, the change magnitudes were computed by the Euclidean distance between two pair vector, and change directions were obtained by comparing the value of pair vector. The magnitude and direction image were computed and the direction of changes were analyzed. Finally, the change image in false color was output. CVA is an effective change detection technique. The accuracy evaluation and direction of change vector analysis need further study.