Tracking using particle and Kalman filters in hand washing quality assessment system

This paper describes a method for tracking the hands/arms of a person performing hand washing. A hand washing quality assessment system needs to know if the hands are joined or separated, if they are under water, if they are in contact with the towel or the tap, and it has to be robust to different lighting conditions, occlusions, reflections and changes in color on the steel surface. In the proposed system hands/arms are extracted by using skin color segmentation. An area based ellipse model is used for representing each hand/arm. A Particle filter (PF) in combination with a k-means based clustering technique is used for tracking both hands/arms. A supervision algorithm measures the number of objects being tracked and the quality of the tracking itself. Finally the PF performance is discussed and compared with the standard Kalman filter (KF) estimator.