A dual approach for quantitative gait analysis based on vision and wearable pressure systems

Abstract

This paper proposes two approaches to characterize gait taking into account only quantitative measurements of dynamic nature. A pair of wireless sensorized insoles are used to obtain gait phases based on the involved forces, and a computer vision system externally estimates measurements through movement analysis. The wearable approach is composed of a pair of insoles, consisting of an assembly of FSRs and an inertial measurement unit. A micro-controller provides the captured data to a Bluetooth module that transmits it to be processed. The vision system obtains gait features using a single RGB camera. We have developed an algorithm to extract the silhouette using background subtraction, and locating heel and toe of each foot using the shape of the silhouettes. Detection of Heel-strike and Toe-off is based on gradient. Gait phases and other spatio-temporal parameters are derived from them.