Seamlessly Extracting Simple Gait Parameters by Using RGB Camera Mounted on Rollator

The demand for gait parameters collection for analysis in medical settings has been increasing in recent years due to the aging population. However, the available devices required for such collection are expensive due to RGBD sensors which should be robust to sunlight and require medical personnel to operate. Moreover, current vision systems generally have an implicit assumption that the user's legs are observed clearly. It is a limiting factor for user's life, because the user may want to wear a long skirt. Hence, in this paper, we propose a novel and low-cost solution for the problem, as a visual add-on for standard rollators which only uses the motion of the feet. This add-on device, without constant attention from medical personnel, can provide important measurements for health monitoring, such as the number of steps, cadence, time spent between each step. The device, which mainly consists of an RGB camera and a computational unit, makes use of the algorithm to capture the dense optical flow of only the forward motion of the feet. From this filtered optical flow, the active pixels are counted and inputed into a peak detector to determine the actual steps which occur. Other parameters would be determined based on the steps detected, consequently. The experimental results showed that the device can capture the gait parameters in real time and within reasonable error rate.