Among different remote sensing applications, change detection deserves specific consideration. The importance of this area is its applicability on damage assessment after natural disasters. Fortunately, recent sensors allow researchers to develop advanced change detection methods. Some of these benefit from panchromatic or multispectral remote sensing images, whereas others use 3D data besides the 2D information. In this study, we benefit from both 2D and 3D data to detect changes in buildings. We specifically focused on building change detection, since after a natural disaster damaged building information is one of the most important one. Our building change detection method is based on our previous study based on probabilistic building detection. In this study, we first extract corner points using the Harris corner detector from panchromatic images. These corner points are used on Digital Surface Model (DSM) data to estimate possible building locations. To do so, we represent possible building locations via a kernel based density estimation method. In this study, we use the difference of the bitemporal estimated kernel maps (obtained in two different times) for change detection. Then, we apply a morphology based shape refinement method. As a result, we can detect changes in the scene. We tested our method on WorldView-2 sensor images with 780 buildings. The results are promising.