Depth map-based human activity tracking and recognition using body joints features and Self-Organized Map

Abstract

In this paper, we implement human activity tracking and recognition system utilizing body joints features using depth maps. During HAR settings, depth maps are processed to track human silhouettes by considering temporal continuity constraints of human motion information and compute centroids for each activity based on contour generation. In body joints features, depth silhouettes are computed first through geodesic distance to identify anatomical landmarks which produce joint points information from specific body parts. Then, body joints are processed to produce centroid distance features and key joints distance features. Finally, Self-Organized Map (SOM) is employed to train and recognize different human activities from the features. Experimental results show that body joints features achieved high recognition rate over the conventional features. The proposed system should be applicable as e-healthcare systems for monitoring elderly people, surveillance systems for observing pedestrian traffic areas and indoor environment systems which recognize activities of multiple users.