Online Finger Motion Recognition with Ultrasound Image of the Forearm

Le résumé fourni par la source

In recent years, ultrasound imaging (US imaging) has been gradually used to build a realistic human-machine interface (HMI). However, few studies concern to the real time performance of this type of HMI. In this paper, the primary objective is to systematically evaluate the feasibility of B mode ultrasound image as a real time HMI. Firstly, an online finger motion recognition experiment including six able-bodied subjects was performed. Then, the real time performance was evaluated with four different metrics (motion selection time (ST), motion completion time (CT), motion completion rate (CR), and motion real time accuracy (RA)). Finally, the evaluated results demonstrate that, the ST and CT, which are achieved by the B mode ultrasound based HMI, can meet the requirement of the online prosthetic control. In particular, it is also found that, this type of HMI has demonstrated the excellent performance of the CR and RA for the fine finger motion recognition. Therefore, it is concluded that, the B mode ultrasound based HMI is feasible and can thus be suggested for prosthetic control.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.