Font Size: 

This paper discusses a feature extraction process from a CCTV video using Histogram of Oriented Gradient (HOG) method. The CCTV video contains the face of an individual. The accuracy of the individual recognition system based on CCTV video is strongly affected by the feature vectors. The objective is that based on several videos or frames, the feature vectors are similar for the same individual and the feature vectors are different for different individual. From our experiments, we have the following two conclusions. Firstly, the feature vectors generated from 3 videos containing one individual have similar patterns. Secondly, the feature vectors generated from one video containing 3 different individuals have different pattern (or low similarity) for each individual.

feature extraction; feature vectors; histogram of oriented gradients; CCTV video; individual recognition systems