Human Recognition Improvement Using Fusion Of Gait And Facial Features

Abstract

Biometrics are unique human traits that are stable over time. Face, iris, fingerprint and type of gait are biometrics used in identification systems. Face identification in distant images is one of the challenges that remains to be elucidated due to the lack of face detail, and efforts have been made to address it. One way to overcome this challenge is to use other biometricsbeside the face. In the fusion of biometrics, two or more biometrics are mainly used for identification and improving evaluation parameters. This paper attempts to solve the challenge of identification by fusion of the face and gait biometrics in decision level image fusion using distant images. This paper proposes a new method of identification using face and gait biometrics fusion. In the proposed method, the features of the facial images are extracted by means of the scale-invariant feature transform (SIFT) algorithm. Gait biometrics have been used to overcome the challenge of insufficient face image detail. To overcome the challenge of variations in the type of gait, the maximum curvature and histogram of oriented gradients algorithms have been used to extract the features. The obtained features are categorized using three classifiers namely, supporting vector machine (SVM), K-nearest neighbors and random forest in five stages; and finally by a decision level fusion rule, the identification is done. The proposed method is simulated on the ORL face database and the CASIA A gait type. The recognition rate of 99.50 and the precision of 99.80 indicate the superiority of the proposed method.