Vision based Identification of Joint Coordinates for Marker-less Gait Analysis

Gait analysis has a wide application in medical, rehabilitation, geriatric care, biometrics, sports, animation, and many other avenues. However, gait analysis systems require highly sophisticated devices and methods in a laboratory setup under controlled environment. Consequently, sometimes the subjects are not able to display their natural gait pattern. There is thus a need for a system that works in uncontrolled conditions under practical constraints. This paper proposes a new approach for identification of human joints for gait analysis in a markerless setup or environment. The proposed method has been used successfully to determine coordinates of joints (shoulder, hip, left knee, right knee, left ankle and right ankle). The extracted positions of the joints are then compared with those obtained from marker based identification and ground truth. This comparative analysis performed confirms the efficiency of the proposed techniques used in the determination of the joint trajectory. These trajectory can play crucial role in gait related pathology diagnosis.