Information about project titled 'Validation of a model-based image-matching technique for 3D motion reconstruction from uncalibrated 2D videosequences'
Validation of a model-based image-matching technique for 3D motion reconstruction from uncalibrated 2D videosequences
|Details about the project - category||Details about the project - value|
|Project manager:||Tron Krosshaug|
Biomechanical analyses of injury mechanisms are essential for understanding how to prevent injuries. As these situations cannot be reconstructed in the laboratory, particularly interest lies in utilizing the video data optimally. At present the existing methods for analysing injury situations from videotape are unsatisfactory.
The aim of this study is to present and validate a model based image-matching technique for 3D motion reconstruction by use of uncalibrated 2D video sequences.
Method: Different motions, including both high and low impact motions of different complexity will be subjected to motion analyses by the means of: An infrared state-of-the-art optical 3D tracking system (ProReflex, Qualisys Inc.) 2 Amti Force platforms, 3 ordinary video cameras 3D kinetics and kinematics from the recorded motions will be calculated from the ProReflex system and force platforms, which will yield the gold standard for this experiment. The videotapes are digitized, lens corrected, and deinterlaced to achieve an effective frame rate of 50Hz. Motion estimates and kinetic variables will also be derived from the videotapes using a model matching method based on a commercially available 3D animated human model (Poser, CuriousLabs Inc.) In the Poser interface, one video sequence (or two synchronized sequences) is imported into the background for the animation model. The model is then manually fitted to the background pictures.
We will evaluate the matching performed by using one, two or three camera views from different perspectives.
This study introduces a novel method for analysing uncalibrated video recordings from the field (for example recordings of injury situations). This method is potentially of great benefit in the work of sport trauma research and may eventually help us to design methods for the prevention of injuries.
Root Mean Square (RMS) hip and knee flexion/extension angle differences were less than 12° for all the matchings. Estimates for ad-/abduction (<15°) and internal/external rotation (<16°) were less precise. RMS velocity differences up to 0.62 m/s were found for the single camera matchings, but for the triple camera matching the RMS differences were less than 0.13 m/s for each direction.
In conclusion, a new model-based image-matching technique has been developed, that can be used to estimate temporal joint angle histories, velocities and accelerations from uncalibrated video recordings. The kinematic estimates, in particular for COM velocity and acceleration, are clearly better when two or more camera views are available. This method can potentially be used to arrive at more precise descriptions of the mechanisms of sports injuries than what has been possible without elaborate methods for 3D reconstruction from uncalibrated video sequences, e.g. for knee injuries.