Humans are able to recognize small number of people they know well by the way they walk. This ability represents basic motivation for using human gait as the means for biometric identification. Such biometric can be captured at public places from a distance without subject’s collaboration, awareness and even consent. Although current approaches give encouraging results, we are still far from effective use in real-life applications. In general, methods set various constraints to circumvent the influence of covariate factors like changes of walking speed, view, clothing, footwear, object carrying, that have negative impact on recognition performance.
In this thesis we propose a skeleton model based gait recognition system focusing on modeling gait dynamics and eliminating the influence of subject appearance on recognition. We propose feature fusion scheme for classification and frame based classification which both demonstrate how such system is comparable to other state of the art appearance based methods even in the environments where these have clear advantage. Furthermore, we address the problem of walking speed variation and propose feature space transformation technique that mitigates its negative influence on recognition performance. Together with frame based classification, the method achieves state of the art classification results and outperforms other similar methods.
With the extensive evaluation we demonstrate state of the art performance and robustness of proposed methods under unchanged conditions, varying walking speeds and even under presence of different computer vision related obstacles (e.g. noise, low resolution etc.) that obstruct the performance of real-life computer vision applications in general.
|
