Image processing for remotely teleoperated robotic manipulator system for underwater operations

Abstract

Presents a method for providing the 3-D spatial information of an object targeted by a remotely teleoperated robotic manipulator system for underwater operations. We use two CCD cameras to capture the raw input picture containing the target object. In order to cope with the rapidly changing undersea environment, we apply the fast image segmentation approach that incorporates watersheds analysis and SFT (Synchronous Feature Tuning) merging algorithm to accurately extract the object. In addition, in order to determine the center coordinates of the target object, a red-light laser is used to pinpoint the object, roughly at the center of the object. The red-light laser points on the two pictures are recognized and the corresponding 2-D coordinates of the two points are calculated. Based on the 2-D coordinates thus obtained, we employ the binocular stereo correspondence technique to determine the 3-D coordinates of the object. The experimental result has shown that, at one meter away from the remotely teleoperated robotic manipulator system, an error rate less than 3% can be achieved.