1 August 1990 Kinesis: a model-driven approach to human motion analysis
Author Affiliations +
Proceedings Volume 1395, Close-Range Photogrammetry Meets Machine Vision; 13952R (1990) https://doi.org/10.1117/12.2294342
Event: Close-Range Photogrammetry Meets Machine Vision, 1990, Zurich, Switzerland
Human/biological motion analysis is traditionally addressed as a problem of tracking a set of (active/passive) markers appropriately attached to the body part under study. General, fundamentally three-dimensional movements cannot be studied in this way. After having studied for some time the different aspects of such a subtly complex research topic,1,2 we propose now an approach which is model- driven, in the sense that it is based on an Animated Humanoid and on the comparison between the actual images of the body and the synthetic images of the humanoid. Our system (that we dubbed Kinesis) organizes the computation into three main phases: (1) Image (sequence) acquisition and processing, (2) Humanoid animation, (3) Optimization. Images acquisition and processing is performed on a VDS Eidobrain video- workstation, which allows image sequences to be stored from three video inputs. The second module relies on a distributed model of motor control, which solves motor redundancy through the regularizing properties of muscle elasticity.3,4 The optimization module drives the humanoid animation in such a way to find the best fit between the original data and the humanoid. Kinesis requires two calibration: the photogrammetric calibration of the (at least three) cameras and calibration of the humanoid. Preliminary results are presented.
© (1990) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Pietro Morasso, Pietro Morasso, Massimo Solari, Massimo Solari, } "Kinesis: a model-driven approach to human motion analysis", Proc. SPIE 1395, Close-Range Photogrammetry Meets Machine Vision, 13952R (1 August 1990); doi: 10.1117/12.2294342; https://doi.org/10.1117/12.2294342


Human motion estimation from a single view
Proceedings of SPIE (October 10 2000)
Robust 3D reconstruction system for human jaw modeling
Proceedings of SPIE (March 07 1999)
Recognizing parameterized three-dimensional objects
Proceedings of SPIE (October 02 1994)
An Evolving System For Image Understanding
Proceedings of SPIE (July 21 1985)

Back to Top