A multi-view human bounding volume estimation for posture recognition in elderly monitoring system

In this work, we propose an elderly monitoring system based on a multi-view human posture recognition from overlapping cameras with an application of fall detection. This system is important in order to help elderly people to stay in a secure environment. It classifies person's pose. These postures (standing, crouching, sitting or lying on the floor) is obtained by performing an estimation of the human bounding volume. To obtain this volume, we estimate the height of the person and its surface which is in contact with the ground according to the foreground information of each camera. Using them, we can differentiate lying on floor posture which can be considered as fall to other postures. The foreground information fusion is based on homography projection. For the validation of the proposed algorithm, we have tested it on public fall detection dataset and compare our performance to others state of the art algorithms. The results show the accuracy of our proposed algorithm in issuing fall detection alarm.