REDUCING CHANGE DETECTION ERRRORS DUE TO DIFFERENT POINTINGDIRECTIN OF TIME SERIES DATA

摘要

Different pointing direction of time series data may cause change detection errors.In this study, a change detection method was developed to reduce the change detection errors.For reducing change detection errors, the results of mixture analysis of the time series data wererecalculated to assign values into a grid data. Same areas of the time series data could becompared using the grid data.A simulation was performed to investigate that the proposed method is more reliable forreducing change detection errors than pixel-by-pixel comparison method. The change detectionerrors by proposed method showed maximum 50% less than the result by pixel-by-pixel.As a case study, proposed method was applied to ASTER images. Change detection of themixture analysis results of the two images carried out by proposed method.