Markerless Motion Capture from Single or Multi-Camera Video Sequence

The modeling of human characters consists of two distinguished processes, namely the definition of 3D shape and 3D movement of the body. To achieve realism, both shape and motion information should be acquired from real persons. The two procedures are usually considered separately, by using full body scanners and motion capture systems. In this paper, we report about deterministic methods developed at IGP to perform the two modeling processes by using a unique set of data. Monocular or multi-camera videogrammetry in fact allow for human body modeling and motion reconstruction at the same time. The two reliable and accurate photogrammetric procedures are introduced and examples are presented.