Barrier Detection and Tracking from Parameterized Lidar Data

Barrier detection and tracking is a key technique in augmented reality(AR). By adding up the sense of objects, users are able to safely observe or avoid moving objects in the view. With the development of 3D Lidar technology, the acquisition of 3D points in the scene is getting more efficient. In comparison with image data, the 3D points from Lidar contain reliable depth information in a large range. However, processing unstructured point cloud is less efficient for augmented reality applications. By noticing the structure of the 3D points captured from Lidar, we propose to parameterize the data from Lidar. In this way, we are able to reuse detection and tracking methods from images for a simple solution in barrier detection and tracking from Lidar data. We test our method by using the Lidar data captured from an unmanned ship. The results show that our method can quickly detect the barriers with bounding boxes, indicating the distance, direction and size of the barrier in the scene.