Vision-Based Pedestrian Fall Risk Evaluation

• Main Authors: Wei-Tzu Liu, 劉威澤
• Other Authors: Mau-Tsuen Yang
• Format: Others
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/554634

碩士 === 國立東華大學 === 資訊工程學系 === 102 === 　　Pedestrian fall detections and predictions using wearable sensors incur problems such as inconvenience,inaccuracy,and high cost.Therefore,this thesis proposed a vision‐based pedestrian fall risk evaluation system.Initially,a background subtraction and morphological operations were applied to obtain a region of interest (including a pedestrian). Subsequently, people gait stability,symmetry, and body stability were computed to build a mixed module consisting of seven features.Multiple perspective problem was handled by estimating the angle between pedestrian and the camera.Particularly, three perspectives were considered in our experiments:the 0∘(front), 45∘ and 90∘(side). Because fall risk assessment is a pretty subjective issue, the Analysis Hierarchy Process (AHP) was utilized to determine the weight of each individual feature so that the estimated overall fall risks could approach the mankind judgments.