An Efficient Study for Approaching Biometric with Bio-Inspired Learning Techniques

In recent years, biometric authentication of face recognition system deals with highly challenging and complex problems. When compared with other methods of identification such as iris, speech recognition, gait recognition, fingerprint and palm print, face recognition can efficiently detect a person. Face recognition problems successfully attain a high accuracy rate in age detection, gender recognition, smile detection, and voice detection. However, biometric face recognition does not have an efficient rapid recognition algorithm. On the other hand, computational speed and recognition time play a vital role in the face recognition system. By using the genetic algorithm, one needs to improve existing algorithms in order to achieve high performance based on best matching in face identification. Therefore, this paper introduces an overview of biometric and non-biometric face recognition techniques. Experiments based on FERET database provide a better recognition rate of facial images. The research issues concentrate on facial expressions and partial occlusions in an uncontrolled environment for face recognition. These are deemed as challenging issues, and thus, a solution is proposed to tackle these problems by using a single sample per class. Expected results presented accuracy rate of 99% perfect match compared to the principle approaches.