An automatic laser scanning system for accurate 3D reconstruction of indoor scenes

3D reconstruction of indoor scenes is important for lots of applications like SLAM (Simultaneous Localization and Mapping), VR (virtual reality) and AR (augmented reality) etc. This paper presents a precise and low-cost 3D reconstruction system, which consists of a laser emitter, a monocular camera and a turntable. The system is firstly calibrated to determine the intrinsic and extrinsic parameters of the camera and laser plane. By rotating the scanning device, a serial of laser strip images can be obtained. With the proposed point cloud registration algorithm, multiple 3D profiles of the target scene can be computed precisely and aligned automatically. As a result, a complete 3D model of the indoor scene with color information and high accuracy can be obtained. In the experiments, some real scenes are reconstructed and evaluated from the aspects of both visual quality and measurement accuracy.