A Least-Squares Image Matching Method for Precise 3D Registration of Airborne LiDAR Data and High-Resolution Satellite Images

Abstract

Abstract 3D registration of high-resolution satellite imagery (HRSI) and airborne LiDAR data can be done by the availability of some matched positions for estimating the 3D transforms such as 3D affine or rational functions. The main idea of this paper is the estimation of the 3D registration parameters (here, the 3D Affine) based on the Least-Squares Image Matching (LSIM) method. Traditionally, in the LSIM method, the sum of squares of differences between the matched gray values are minimized by simultaneously estimating a geometric and a radiometric transformation between the matched positions. Although this method is known as a local strategy for matching the perspective images; it has been adapted here as a global strategy for matching between HRSI and LiDAR data. Hence, the proposed method exploits the whole of the overlapped regions between the LiDAR data and the HRSI to find the parameters of a transformation to relate the 3D LiDAR data and the 2D HRSI. Considering the cost function of the LSIM method, the radiometric similarity of the HRSI and LiDAR data has also increased by adding the shadows and topographic effects to the LiDAR intensity data to be act as a proper entity in matching with the HRSI. The LSIM is done through the solving an equation system of nonlinear equations that is sensitive the initial values of the unknowns. The results indicate that the accurate 3D registrations with the one-pixel precisions can be accessible even when the approximate geometric transformations are used as the initial parameters.