The initial study of LLS-based binocular stereo-vision system on underwater 3D image reconstruction in the laboratory

Abstract

This study aims to develop a three-dimensional image reconstruction method based on the Laser Line Scan (LLS) technique to establish the binocular stereo-vision system for the preliminary research of obstacle detection technique of Autonomous Underwater Vehicle (AUV). A coordinate mapping relationship between 2D pixel coordinate and 3D world coordinate, which can be used to reconstruct 3D objects from 2D scan data, is established by means of direct camera calibration in air and water. In the experiments, the target object was originally designed by Computer-Aided Design (CAD) model and fabricated by the 3D printer. Subsequently, the qualities of point clouds acquired from the target object would be analyzed and compared in the stability water tank at National Cheng Kung University. The acquired point clouds would be used for polygonal surface estimation of the target object by Bonjean curve fitting method in water, with the reference results in air. The acquisition of raw point clouds has been accessed via the transformation to grayscale, histogram equalization, image binarization and skeletonization thinning. Consequently, the results evaluated by our stereo-vision system indicate the reliability and performance in the stability water tank before the application to the obstacle-avoidance of the AUV.