Skip to Main Content
This paper describes an approach to analyze the facial appearance of building in image. Detecting the line segments and grouping them make a mesh of parallelograms. The belongings of face as principal component parts (PCPs) as doors, windows, walls are detected by merging neighborhood of parallelograms with similar color. We use MSAC to group such line segments which have a common vanishing point. We calculate one dominant vanishing point for vertical direction and maximally five dominant vanishing points for horizontal direction. The vertical group and one of horizontal groups create a mesh of basic parallelograms. Each mesh represents one face of building. Each face contains windows, doors and the wall. The PCPs are separated into two groups according to their properties. The first one is represented by windows and doors. This group caries geometrical information of building such as how many windows and doors there are. So the robot can measure that how tall and wide the building is. The second one is wall region containing an important visual information such as what color of building is. This information can help the robot to distinguish a building from others. The proposed approach can apply to building detection, recognition and reconstruction.