Information on 3-D structure expands the scope of change detection applications, for example, in urban studies, human activity, and forest monitoring. Current change detection methods do not fully consider the specifics of SAR data or the properties of the corresponding image focusing techniques. We propose a three-stage method complementing the properties of 2-D and 3-D very high-resolution (VHR) synthetic aperture radar imagery to improve the performance of 2-D only approaches. The method takes advantage of back-projection tomography to ease translation of the 2-D location of the targets into their corresponding 3-D location and vice versa. Detection of changes caused by objects with a small vertical extent is based on the corresponding backscatter difference, while changes caused by objects with a large vertical extent are detected with both
backscatter and height difference information combined in a conditional random field. Using multitemporal images, the kappa coefficient improved by a factor of two in comparison with traditional schemes.