3D Surface reconstruction of transparent objects using laser scanning with LTFtF method

Abstract

Three-dimensional (3D) vision plays an important role in robotics and industrial automation, wherein the reconstruction of transparent objects poses a great challenge. In this paper, we present a scheme for reconstructing the 3D surfaces of transparent objects based on a combination of stereo matching and laser scanning. The traditional laser scanning method can achieve good results in the 3D reconstruction of opaque objects. However, this method will fail for transparent and semitransparent objects because both the front and back surfaces of these objects will reflect the feature laser line. It is difficult to determine the reflection surface of the feature lines captured by the camera. To distinguish the laser reflected by the front surface from the distracting laser reflections by the other surfaces, a laser tracking frame to frame (LTFtF) method is proposed. The laser lines reflected from the front surface of the target are then reconstructed using stereo vision. The entire 3D surface is obtained through laser scanning. In addition, we improve the matching accuracy and point cloud density through an internal pixel matching method. Various experiments were conducted to verify the capability of the proposed method, which includes the reconstruction of transparent and semitransparent objects of various shapes. The proposed method was compared with traditional laser scanning algorithms.