الفهرس | Only 14 pages are availabe for public view |
Abstract Gait recognition is receiving growing interest within the computer vision community. Many approaches are introduced and evaluated recently in this scope. Gait recognition systems are still in evolution phase and there is no system yet that can extract all gait features under complicated conditions and different views. In this thesis, we aim to create a new algorithm, using MATLAB technical computing language, to develop the gait system. Gait has a unique advantage over other biometrics systems. It has potential to be used at a distance without subject observation. Based on this, we do not need the person,s details, but only the outer structure. In other words, the person،s silhouette is needed. in this thesis, a new temporal template technique based on wavelet transform is proposed. It can be considered as an evolution of gait energy image. The aim is to create a new method that combines both structure and movement of the body in one template, and the remaining of the movement in another template, which will cover the wide variation on gait. |