Abstract
3D models play a vital role in intelligent transport systems and in geographic survey. Several methods are available for construction of such models. However, the challenge lies when requirement is to build a 3D model in world coordinates in urban environments because of poor availability of GPS signal. We deal with such challenge by capturing GPS data at several positions along with range images at those positions. The range images are aligned in local coordinate system using fast ICP algorithm. GPS data at selected positions with high probability of position estimation accuracy is used to estimate the transformation between GPS position estimates and scan center positions obtained from simultaneous alignment of range images. The obtained transformation is used to transform all the aligned 3D points into world coordinates. To evaluate the correctness of constructed model, various evaluation methods based on orientation of building facades, positioning on aerial photos and overlapping of neighboring regions are proposed. The evaluation results confirm the correctness of the global alignment.[This abstract is not included in the PDF]
