Automatic Orientation Estimation of Multiple Images with Respect to Laser Data

In this paper, we present an automatic algorithm to estimate the orientation of multiple images with respect to terrestrial laser data. The proposed algorithm takes advantage of both conventional single-view and multi-view registration approaches. The algorithm consists of three steps. In the first step, intensity images are generated from the terrestrial laser data with intensity information, and the initial exterior orientation parameters (EOPs) of each camera image are then estimated from feature correspondences with the intensity images. In the second step, the initial EOPs obtained from previous step are refined through a bundle adjustment process. In the third step, the point cloud reconstructed from the bundle adjustment process is registered to the terrestrial laser point cloud to determine the final orientation of each image. The proposed algorithm is tested on experimental datasets, and the results demonstrate that the proposed algorithm can efficiently handle the orientation estimation of multiple images with respect to the terrestrial laser data.