Target Locating Research based on Stereo Vision for Underwater Vehicle-Manipulator System

P. Gong, Q. Zhang and A. Zhang (PRC)


Underwater vehicle-manipulator system, target locating, CLAHS, SURF


The autonomous underwater vehicle-manipulator system (UVMS) is a kind of UVMS which can have autonomous operation abilities for underwater application. It employs a combination of sensors to autonomously detect targets, conduct environmental measurement and finally implement operations with its manipulator. In this paper, a target locating method based on stereo vision is proposed for trajectory planning of autonomous UVMS. First, the raw images are captured by a binocular camera and preprocessed by the contrast limited adaptive histogram specification (CLAHS) algorithm to make contrast enhancement. Then SURF features are extracted and tracked, and the 3D point-coordinates are calculated by the triangulation method. Two methods, the homography transformation method and geometric centroids method, are proposed to estimate the target’s position under the camera frame. Finally the location estimation result is analyzed under the camera’s frame and the world frame, and the errors on three directions of the world frame are given. Experiments indicate that this method is effective and easy to use while supplying the aid to UVMS for locating the target.

Important Links:

Go Back