A method of 3D scene reconstruction with vehicle-borne laser scanner based on monocular visual positioning

Abstract

In the process of 3D reconstruction of the target scene with the moving vehicle-borne single-line scanner, a method of monocular visual positioning is used to locate the exact position of the moving scanner. Attitude sensor, camera and the laser scanner are located on the same platform which is installed on the vehicle. Data can be collected synchronously. We obtained a top view according to the observation of camera's attitude by IPM. A series of planform are acquired at different moment. After dealing with these pictures with operations of extraction Speeded Up Robust Features (SURF), matching characteristic points and RANSAC algorism, we calculated the transformation matrix between two adjacent vertical views. The position and direction of the scanner at this moment can be calculated. On this basis, depending on the sensors' synchronization control algorithm, the original point clouds data got by laser scanner can be registered as a reference to the traveling track of the moving scanner, and then the 3D reconstruction of the target scene is established. Experiment result shows that this method is easy to operate, time-efficient, low cost, and the accuracy of the method in reconstructing 3D scene is demonstrated.