Automatic Building Extraction in Urban Areas Using a Modified Active Contour Model

Automatic extraction of objects from data and images has been a topic of research for decades. This paper proposes an improved active contour model focusing on building extraction from color aerial images and LiDAR data. An active contour is defined as an energy minimizing spline guided by external constraint forces and influenced by image forces that pull it toward features such as lines or edges. Based on the radiometric and geometric behaviors of buildings, the active contour model is modified in two aspects: the criteria for the selection of initial seeds and the external energy function. The proposed active contour model associated the new height similarity energy factor and regional similarity energy as well as Gradient Vector Flow (GVF) attract the active contour approaching the object contours efficiently. Compared with traditional active contour model, this algorithm can converge to the true building contours quicker and more stable especially in complex urban environment. Investigation of results expressed that, buildings extracted from complex suburban area have 96% of shape accuracy, 95.9% of overall accuracy and 84% k-Factor.