Underwater 3-D imaging and detection by laser plane scanning

In order to satisfy the requirements of the high precision detection for autonomous underwater vehicles (AUV) at close range, and considering the difficulty and inaccuracy of reconstructing 3-D shape from blur underwater intensity image, a range image method based on active triangulation is presented to use green plane laser as the light source to scan submerged targets, and use CCD camera to capture the laser stripe, then the 3-D coordinates of targets can be obtained through digital image processing and integrating the information of position and attitude given by INS. In this paper, the mathematical model of measuring system and the influence of water-air interface refraction are specially discussed, and the configuration parameters are pre-calibrated with mechanisms through positioning the intersection of the optical axis and the laser plane. Furthermore, an algorithm for laser stripe detection is introduced, in which the combination of thresholds and the detection template lessens data throughput with 3 orders of magnitude (360x360pixels). Experimental results show that the resolution is approximately 1mm in 3-D coordinates in clear water at 0.7-1.5m ranges; the maximal relative error is less than 1%, and Laser Plane Scanning is a complement method of sonar for underwater detection.