Shape-based Change Detection and Information Mining in Remote Sensing

Change detection is an important application of remote sensing. This paper presents our approach to retrieve and represent interesting shapes in the remotely sensed imagery using supervised classification, edge detection, and polygonal approximation techniques, and to compare the shape similarity by a computationally efficient metric. The experiments were conducted on a time series of calibrated and registered Landsat MSS images, covering the scene of the lakes at the western Nebraska. The results show the effectiveness of the shape-based change detection approach, which is potentially useful for specific applications such as the study of the lake change response to short or long term climatic variation and flood or drought monitoring.