Abstract:
Gait is a particular way or manner of moving on foot.Gait recognition is identifying a personby the manner of its walk.This is a marker less and unobtrusive biometric; offering the possibility to identify people from a distance and without interacting with them; this property makes it an attractive method for identification. The project aimedat developing a system capable of automatic gait recognition. A person’sgait signature was created using a model based approach. Temporal and spatial metricsextracted from the model fitted to the individual; such as changein angles of the limb or the stride length of a person’sgaitwere used to create a “gait signature” of the individual which weretransformed in Eigen Space using Principle Component Analysis and were later used to identifythe subject in videosegments. The project provided promising results with the technique of Principle Component Analysis and even with the self-similarity plots. Limb angles proved the best way to extract a gaitsignature and angular velocities showed quite degraded results.Principle Component Analysis offered a good way to represent most of the variation in the
data, whilereducing dimensionality. Recognition rates up to 80% were achievedby the project, further strengthening the notion that gait can be used as a biometric.
