Finger Localization and Classification in Images based on Generalized Hough Transform and Probabilistic Models

Modern sliding table saws are equipped with several passive safety devices. However, alone in Germany several hundred severe injuries or limb amputations occur each year. Thus an active safety system is developed in order to protect the machine user. For this purpose, a camera is applied as the main sensor system for hazard analysis. When a dangerous situation is recognized, the system triggers a rapid saw blade braking device. The hazard analysis is based on a localization and classification step. For this, the user´s hand is represented by a probabilistic model relating to a feature extraction utilizing the generalized Hough transform. The described procedure works reliably, however, the computational complexity is demanding