Change detection in very high resolution imagery based on dynamic time warping: An implementation for Haiti earthquake damage assessment

In this paper, we develop an approach for change detec tion in VHR imagery by using the discrete cosine transform (DCT) and dynamic time warping (DTW). The approach considers both contextual information and more accurate registration of two images. The proposed approach was evaluated by using panchromatic images with 0.5 spatial resolution taken from Worldview satellite, and Haiti earthquake-induced changes were detected. A small region was selected for analysis where the Haiti palace was located, and the change map was visually compared to the change map obtained by using standard change detection tool of Erdas Imagine.