作者: S.S. Intille , J.W. Davis , A.F. Bobick
关键词: Feature (computer vision) 、 Track (disk drive) 、 Cluster analysis 、 Computer vision 、 Algorithm design 、 Computer science 、 Tracking (particle physics) 、 Motion estimation 、 Cognitive neuroscience of visual object recognition 、 Object (computer science) 、 Artificial intelligence
摘要: A real-time tracking algorithm that uses contextual information is described. The method capable of simultaneously multiple, non-rigid objects when erratic movement and object collisions are common. closed-world assumption used to adaptively select weight image features for correspondence. Results testing the limitations discussed. has been track children in an interactive, narrative playspace.