Elastic Image Registration for Landslides Monitoring

Landslide is a type of mass movement that causes damage in many areas. The evolving remote sensing technology in producing high resolution images may help in landslide studies. However, the problem in detecting small size landslides is still challenging when suitable image resolution of the area being analyzed is not available. In this paper, a novel method based on elastic image registration, appropriate for the detection of small landslides will be presented. This method can be used to detect and quantify landslide movement with sub-pixel accuracy. It is based on the invocation of deformation operators which imitate the deformations expected to be observed when a landslide occurs. The similarity between two images is measured by a similarity function which takes into consideration grey level value correlation and geometric deformation. The geometric deformation term ensures that the minimum necessary deformation compatible with the two images is employed. An extra term, ensuring maximum overlap between the two images is also incorporated. There are two versions of this method. One using the correlation coefficient as a measure of similarity for the grey level value, and another one using mutual information. These methods are tested using known small scale landslides images of southern Italy taken from the Landsat 5 TM. The mutual information-based method gives more reliable results.