ABSTRACT:This work presents a method that automatically detects, analyses and then updates changes in LiDAR point clouds for accurate 3D urban cartography. In the proposed method, the 3D point cloud obtained in each passage is first classified into 2 main object classes: Permanent and Temporary. The Temporary objects are then removed from the 3D point cloud to leave behind a perforated 3D point cloud of the urban scene. These perforated 3D point clouds obtained from different passages (in the same place) at different days and times are then matched together to complete the 3D urban landscape by incremental updating. Different natural or man-made changes occurring in the urban landscape over this period of time are detected and analyzed using cognitive functions of similarity and the resulting 3D cartography is progressively modified and updated accordingly. The results, evaluated on real data using different standard evaluation metrics, not only demonstrate the efficacy of the proposed method but also shows that this method is easily applicable and well scalable, making it suitable for handling large urban scenes.
INTRODUCTIONIn this paper we present a new method for 3D urban cartography that automatically detects different man-made or natural changes occurring in urban environment and then effectively incorporates them in the resulting 3D point cloud of the cartography by incremental updating. Automatic detection of changes in urban environment for updating 3D urban models, cartography and maps recently become a hot topic in the scientific community as it continues to offer several challenges (Champion and Jurgen, 2009). Most of the proposed techniques detect changes in the urban environment from airborne data using DSMs (Digital Surface Models) i.e. (Vu et al., 2004). Vögtle and Steinle (Vögtle and Steinle, 2004) propose a methodology for detecting changes in urban areas following disastrous events. Instead of solely computing the difference between the laser-based DSMs, a region growing segmentation procedure is used to separate the objects and detect the buildings; only then, an object-based comparison is applied. But this method remains susceptible to misclassifications. Bouziani et al. (Bouziani et al., 2010) presented a knowledge-based change detection method for the detection of demolished and new buildings from very high resolution satellite images. Different object properties, including possible transitions and contextual relationships between object classes, were taken into account. Map data were used to determine processing parameters and to learn object properties. Unlike these methods, in our work we handle the 3D point cloud at 3D grid level for change detection, instead of object level, making it more robust. Most of the work using terrestrial laser scans focuses on deformation analysis for designated objects. Change is detected by subtraction of a re-sampled set of the data (Schäfer et al., 2004), or adjustment to surface models like planes and cylinders (Van Gosliga et al., 2006). In order to d...