Gait recognition by dynamic cues

Abstract

Many studies have now shown that it is possible to recognize people by the way they walk. As yet there has been little formal study of people recognition using the kinematic-related gait features. We present a new method for gait recognition using dynamic features including the angular measurements of the lower limbs as well as the spatial displacement of the trunk. Gait signatures are derived using a feature selection algorithm which is based on a validation-criterion. We show that gait angular measurements derived from the joint motions mainly the ankle, knee and hip angles, possess most of the discriminatory potency for gait recognition with an achieved correct classification rate of 95.7%.

Matched-Pair Classification Analysis on Siblings' Gait Biometric Data

“…By tracking each person's lower limbs from a sideway's point of view, the displacement values of the lower limbs with respect to his/her body's vertical axis throughout the video have been chosen as that person's gait feature, as illustrated in Figure 1(a). This follows successful similar research on unique individuals [12][13][14][15]. The forward movement of the lower limbs gives positive degree values and the reverse is a negative degree.…”

Human Gait identification System

“…Recently, the using of gait as a biometric in human identifications and recognitions has been widely considered by researchers [5], proved that the walking behavior angular measurements obtained from joint movement fundamental hip angle, knee angle, ankle angle, have most of recognition power for walking behavior identification with a completed valid classification rate of 95.7%. In [6], authors proposed technique for walking behavior identification, which was depended on estimated joint angle.…”

Gait recognition by combining wavelets and geometrical features

“…Worapan Kusakunniran, Qiang Wu, Hongdong Li, and Jian Zhang [8] propose constructing binary pattern from a sequence of aligned silhouettes and then adaptive weighting technique was applied to discriminate significances of the bits in gait signatures. Imed Bouchrika and Mark S. Nixon [11] used dynamic features including the angular measurements of the lower limbs as well as the spatial displacement of the trunk for the gait feature extraction. Hu Ng, Wooi-Haw Tan, Hau-Lee Tong, Junaidi Abdullah and Ryoichi Komiya [13] divided the silhouette into six body segments based on anatomical knowledge and then applied Hough transform to obtain the joint angles from the body segment skeletons.…”