Detection, localization and recognition of objects and nearby obstacles are critical in employing intelligent (autonomous) underwater vehicles for surveillance, collision- free navigation, target tracking, docking, surveying and mapping the ocean floor. Although deployment of video cameras onboard submersibles have become commonplace, computer processing of underwater images for scene reconstruction and interpretation is a technology yet to be developed. This paper describes steps toward the development of an active vision system for intelligent data acquisition and processing. This provides the capability to obtain information about the scene from video images online, and to exploit this knowledge in enhancing the data acquisition process; by achieving optimum parameters and geometry of the imaging system to improve image quality and information content.
展开▼
