Comparison Of Gait-Face Fused Human Recognition Techniques
In this paper we have proposed two different techniques for developing a reliable Human Recognition System.
The model is developed with the help of real time Open CV and Emgu CV library. In our approach we have made use of two
different techniques i.e., Model Based Technique and Holistic Based Technique for extracting two different sets of features.
Model Based Technique is used for extracting certain geometric distances related to static and dynamic features from the
Face and Gait Biometrics. Holistic based Technique is used for extracting certain statistical features related to the whole
image of the person. Finally the feature sets obtained for Face and Gait are fused together using three alternative
classification techniques (Bayes / Artificial Neural Networks / Support Vector Machines) and the relative accuracy of these
two techniques is tested. Once recognition is done, automatically attendance will be updated in an Excel Sheet along with his
photo, name, date and time. Our system can automatically update the Database for the newly enrolled persons. Since gait is
captured from a distance, for the convenience of the user we have integrated hand gesture recognition to our system, with the
help of which we can use our hand as a virtual mouse and control the GUI from larger distances. We have an another option
of controlling the GUI by giving voice instructions to it (eliminating the need of a mouse interface) and our system can
speak back to us.