Abstract:
Nowadays, Gait is a crucial field for many Pattern Recognition researchers. It is
considered as a good way for biometric authentication in many surveillance systems.
The most important issue in gait recognition is the features extraction from the
silhouettes which are converted from walkers' images or videos. In this paper a new
method has been introduced to identify night walker images, which are captured by
infrared cameras. The new method depends on the silhouette's presence percentage on
different horizontal levels. Minimum distance classifier has been used to choose the
active horizontal levels that lead to the highest identification rate. A proposed algorithm
which is used to select the Active Horizontal Levels (AHLs) has been presented. The
proposed method was evaluated against CASIA silhouette database, to recognize walker
into one, or multi silhouettes. The experimental results reveal the effectiveness of our
new method against other Gait recognition methods to achieve a better recognition rate.