THE EFFECTS OF IMAGE MISREGISTRATION ON THE ACCURACY OF REMOTELY-SENSED CHANGE DETECTION

Image misregistration has become one of the significant bottlenecks fo r improving the accuracy of multisource data analysis, such as data fu sion and change detection. In this paper, the effects of misregistrati on on the accuracy of remotely sensed change detection were systematic ally investigated and quantitatively evaluated. This simulation resear ch focused on two interconnected components. In the first component, t he statistical properties of the multispectral difference images were evaluated using semivariograms when multitemporal images were progress ively misregistered against themselves and each other to investigate t he band, temporal, and spatial frequency sensitivities of change detec tion to image misregistration, In the second component, the ellipsoida l change detection technique, based on the Mahalanobis distance of mul tispectral difference images, was proposed and used to progressively d etect the land cover transitions at each misregistration stage for eac h pair of multitemporal images. The impact of misregistration on chang e detection was then evaluated in terms of the accuracy of change dete ction using the output from the ellipsoidal change detector. The exper imental results using Landsat Thematic Mapper (TM) imagery are present ed. It is interesting to notice that, among the seven TM bands, band 4 (near-infrared channel) is the most sensitive to misregistration when change detection is concerned. The results from false change analysis indicate a substantial degradation in the accuracy of remotely sensed change detection due to misregistration, It is shown that a registrat ion accuracy of less than one-fifth of a pixel is required to achieve a change detection error of less than 10%.