作者: Martin D. Levine , Douglas A. O'Handley , Gary M. Yagi
DOI: 10.1016/0146-664X(73)90024-5
关键词: Computer graphics (images) 、 Computer science 、 Computer vision 、 Context (language use) 、 Artificial intelligence 、 Window (computing) 、 Heuristic (computer science) 、 Depth map 、 Professional video camera 、 Correspondence problem 、 Stereoscopy 、 Robot
摘要: The research in support of the integrated robot project Lit Jet Propulsion Laboratory is partially directed towards problem visual perception by computer. It anticipated that autonomous behavior will be predicated on feedback obtained an analysis environment. Thus equipped with a pair identical television cameras which provide digitized stereoscopic input. Using latter, one approach to representing three-dimensional objects scene depth map. This paper describes computer method for obtaining map mirrors detail original scene. To this end, adaptive correlation window incorporated as aid solution correspondence problem. Heuristic strategies based local context and texture measures have also been invoked. Results are included demonstrate success technique.