Change Detection for Map Updating with Classification Posterior Probability of HJ Image and TM Image

Change detection with different sensors images is difficult generally. In most cases post-classification change detection method can only be used in which classification errors accumulation lead to many fault detected changes. In this paper, based on the need analysis of change detection for map updating and characteristic analysis of HJ image and TM image, a combined change detection method, change vector analysis with classification posterior probability, was used to get changed geographic information. After independent classification of HJ image and TM image, the posterior probabilities of classification were applied to change vector analysis to get change types and change probability. This method integrated the advantages of change vector analysis detection and post-classification detection, in which errors of change detect were reduced What's more, when classification posterior probabilities were used in change detection of geographic features, not only "from-to" type of change but also change probability can be got. The later was obtained through a special changed thresholds setting. Results analysis showed that the method could get good change detection accuracy using HJ image and TM image, which provided a feasible way for change detection in map updating.