Project Group: Cvg10-1025 Participants:

This project explores the possibility of creating a non obtrusive and intuitive gesture interface for wearable computers. A computer vision solution based on the CONDENSATION algorithm is proposed, and the underlying tracking algorithms designed and implemented. A new measurement model and contour likelihood function based on both the shape and color of the hand is presented. The hand tracking system is tested on scenes with varying lighting, motion blur, cluttered background and other skin colored objects. Even under these difficult conditions the performance is concluded to be good enough for wearable interface purposes. The report is structured as follows: The Introduction introduces the problem, In the Analysis chapter the theoretical aspects and developed algorithms are described. Design explains the specific adaptations of the algorithms to the task of hand tracking. Then Results And Discussion deals with the presentation and interpretation of the findings respectively. Equations, figures and tables are enumerated successively within each chapter. The bibliography can be found in the back of the report. Bibliographic citations are marked as (Author[year]).