Building change detection using multi-sensor and multi-view- angle imagery

Abstract

Change detection of buildings in urban areas is very challenging due to geometric distortions in very high resolution (VHR). These distortions create problems in the co-registration of different images. Thus, it is very problematic to exploit images acquired by different sensors and different view angles using conventional change detection methods. Therefore, the majority of studies in this field avoid using multi-sensor and multi-view angle images. In this study, a novel co-registration method, called Patch-Wise Co-Registration (PWCR), is used to contribute to a solution of the problem. This method integrates the sensor model parameters into the co-registration process to relate corresponding pixels. From the corresponding pixels, corresponding segments (patches) are generated. Later on, the brightness values of the matching pixels/segments are compared in order to detect changes. Here, a Multivariate Alteration Detection (MAD) transform is used for identifying the changed segments. The proposed method provides the opportunity to utilize various images as bitemporal sets for change detection.