Gait Analysis for Recognition and Classiﬁcation
The Problem: Human gait is an identifying feature of a person that is determined by his/her weight, limb length, and habitual posture. Hence, gait can be used as a biometric measure to recognize known persons and classify unknown subjects. Gait can be detected and measured at low resolution, and therefore it can be used in situations where face or iris information is not available in high enough resolution for recognition. We have designed a representation for the dynamics of human gait that facilitates the recognition and classiﬁcation of people by their gait.
Motivation: Gait is deﬁned as “a manner of walking” in the Webster Collegiate Dictionary. We extend our deﬁnition of gait to include both the appearance and the dynamics of human walking motion. had shown in the 1970’s that observers could recognize walking subjects familiar to them by just watching video sequences of lights afﬁxed to joints of the walker. Hence, in theory, joint angles are sufﬁcient for recognition of people by their gait. However, recovering joint angles from a video of walking person is an unsolved problem. In addition, using only joint angles ignores the appearance traits that are associated with individuals, such as heavy-set vs. slim, long hair vs. bald, and particular objects that one always wears. For these reasons, we have included appearance as part of our gait recognition features. Previous Work: Given the ability of humans to identify persons and classify gender by the gait of a walking subject, there have been a few computer vision algorithms developed for people identiﬁcation and activity classi- ﬁcation. Cutler and Davis used self-correlation of moving foreground objects to distinguish walking humans from other moving objects such as cars. Polana and Nelson detected periodicity in optical ﬂow and used these to recognize activities such as frogs jumping and human walking. Bobick used a time delayed motion template to classify activities. Little and Boyd used moment features and periodicity of foreground silhouettes and optical ﬂow to identify walkers.