Remote sensing image thresholding methods for determining landslide activity

Abstract

Detecting landslides and monitoring their activity is of great relevance for disaster prevention, preparedness and mitigation in hilly areas. To this end, change detection techniques were developed and applied to multi‐temporal digital aerial photographs, simulating the very high spatial resolution of new satellite sensor optical imagery, over the Tessina complex landslide in north‐eastern Italy. Several automatic thresholding algorithms are compared on the difference orthorectified and radiometrically normalized images, including some standard methods based on clustering, statistics, moments and entropy, as well as some more novel techniques previously developed by the authors. In addition, a variety of filters were employed to eliminate much of the undesirable residual clutter in the thresholded difference image, mainly as a result of natural vegetation and man‐made land cover changes. These filters are based on shape and size properties of the connected sets of pixels in the threshold maps. This has enabled us to discriminate most ground surface changes related to the movement of a pre‐existing landslide.