Constrained LMDS technique for human motion and gesture estimation

Abstract

Body Area Networks is an emerging domain taking a big interest from developers and system designers. On the other hand, the need to localize is becoming necessary in diverse applications. Within this context, the aim of this paper is to estimate the different gestures and motions of the human body. Initially, we use information, about human motion, extracted from C3D files. In fact, these files provide us with the exact 3D coordinates of the sensors on a moving body. In a second step the IEEE 802.15.6 channel model is used to estimate the distances between sensors which are the input of the locomotion technique based on Multidimensional Scaling. Basically, this technique did not present satisfying results, that's why we have improved our results by an SVD reconstruction algorithm and by adding distance constraints.