A Two-step Displacement Correction Algorithm for Registration of Lidar Point Clouds and Aerial Images without Orientation Parameters

Abstract

This paper presents a novel linear registration algorithm for lidar point clouds and aerial images without orientation parameters. First, preprocessing is conducted to classify the lidar point clouds into ground points, building points, and aboveground, non-building points. After preprocessing, the algorithm consists of two sequential steps, i.e., Tilt Displacement Correction and Height Displacement Correction. As the kernel of the proposed registration algorithm, the mathematical model for Height Displacement Correction is a set of linear formulas analytically deduced from the rigorous geometric function of a single image. The proposed registration algorithm does not require any orientation parameters for the image, which greatly lowers the requirements for image acquisition. Due to the model's linearity, the proposed algorithm is computationally efficient. Our experimental results demonstrate that the proposed algorithm can register aerial images vvithout orientation parameters at the same accuracy level of space resection based on collinear equations. This result fulfills the requirement for the fusion of lidar range data and aerial images in most large-scale urban modeling applications.