We explored surveillance solutions based on gait recognition, i.e., the identification of people based on their walking style. Compared to other biometrics, gait has some unique characteristics. The most attractive feature of gait as a biometric trait is its unobtrusiveness, i.e., the fact that, unlike other biometrics, it can be captured at a distance. Although the study of kinesiological parameters that define human gait can form a basis for identification, there are apparent limitations in gait capturing that make it extremely difficult to identify and record all parameters that affect gait. In practice, gait recognition systems have to rely on a video sequence that is taken in uncontrolled environments.
When capturing conditions cannot be controlled, a variety of problems arise due to occlusions, varying illumination levels, cluttered backgrounds, or changing walking directions. We developed methodologies that dealt with these problems and delivered superior recognition performance, exceeding 95%, without relying on texture or facial information.
Our systems relied both on model-based approaches, where a human model was the basis for gait modelling, and holistic techniques, where the shape of human silhouettes was the only information used.