Human Identification Based on Gait

Analysis of person’s behavioral and psychological features like Gait or the manner of walking, allows human identification process to be able to extract the gait information from a video sequence of different persons at a distance and recognize the specific person. This paper proposes a fast gait recognition algorithm based on the averaged silhouette of person. Three important novelties in our proposed algorithm cause that it gives a better performance than previous methods in both recognition rate and computational speed.۱- frame sampling method, which causes the algorithm computational time, without any change in recognition rate, decrease to about half. ۲- New method for background estimation, which gives an acceptable estimate of the background. ۳- Local thresholding method, which gives more complete and more accurate binary silhouette pictures. Experimental results demonstrate that our proposed algorithm accuracy is superior to other previous methods in the gait recognition application.

Keywords:

Biometric Recognition , en , Gait Recognition , Background subtraction , Averaged silhouettes