Human identification system based on feature level fusion using face and gait biometrics

During the past years, face and gait recognition in video have received significant attention. Consequently, their recognition problems have challenged due to largely varying appearances and highly complex pattern distributions. However, the complementary properties of these two biometrics suggest fusion of them. Face recognition is more reliable when the person is close to the camera. On the other hand, gait is a suitable biometric trait for human recognition at a distance. Information from these two biometric sources, frontal of face and side of gait, are utilized and integrated at feature level. Face image is represented by the Active Lines among Face Landmark Points (ALFLP) feature vector. Gait image is represented by the Active Horizontal Levels (AHL) feature vector. Face and gait feature vectors are fused using a proposed effective fusion method. The proposed system was tested on CASIA database and the achieved results showed that the integrated face and gait features carry the most discriminating power compared to any individual biometric.