Literature Survey on Building 3-D Models of underwater Objects

摘要

Three-dimensional object recognition is a topic of active interest motivated by a desire to build computers with “humanlike” visual capabilities, and also by a pragmatic need to aid numerous real world applications such as robot bin-picking, autonomous navigation, automated visual inspection and parts assembly tasks. Building 3D models from multiple image sequences is a classical but challenging area in computer vision and computer graphics research. We propose a 3D modeling method that drastically improves robustness over existing approaches, and explicitly deals with underwater objects. Generated 3D models can be used to represent 3D objects in the first stage of underwater 3D target recognition system for Autonomous Underwater Vehicles (AUVs).