FACE AND VOICE RECOGNITION BASED BIOMETRIC SYSTEM DESIGN
Project no. 722 · Status: Kapanmış · 2016
Summary
With modern life and increasing technology, the need for security measures is increasing day by day. With increasing security, interest in biometric security systems such as face and voice recognition is increasing. In the first part of the study, voiced regions of speech were detected by using zero-crossing rate and short-time energy, different features were extracted from voiced regions and a combination of features giving the best success rate was found by experimenting with different kernel functions in support vector machine classifier. In the second part of the study, the facial region was determined by Viola-Jones algorithm, different features were extracted from the facial regions and a combination of features giving the best success rate was found by experimenting with different kernel functions in support vector machine classifier. The experiments were conducted first in the ready-made databases and then in the database created by the team. In the database created by the team, 99% success rate in voice recognition and 100% success rate in face recognition were achieved. The increase in the number of subjects caused a decrease in the performance of the system.