Low-complexity registration of visual imagery with 3-D LiDAR

A robust 3-D model reconstruction plays vital role in many real world applications like city planning, architecture design and cartography. In this paper, we develop a low complexity robust hybrid model-based algorithm for registering oblique aerial images onto a 3D model generated from airborne Light Detection and Ranging data. Our proposed algorithm consists of two steps. In the first step, we use an intelligent search process to roughly estimate the extrinsic parameters of a calibrated camera without any apriori knowledge. The estimated camera parameters are used to transform the Digital Surface Model (DSM) to a view close to the view from which the visual image is taken. The second step is an adjustment process, in which we determine the transformation that maps the visual image onto DSM. Experiments were conducted on a real world datasets.