The Research On Canonical View Synthesis And Gait Energy Image For Human Identification

In recent years identification of individuals using biometrics has gained growing research interests. Biometric is used in a wide array of applications, which makes a precise definition difficult to establish. Biometrics comprise methods for uniquely recognizing humans based upon one or more intrinsic physical or behavioral traits. Biometric characteristics can be divided in two main classes:Physiological which is related to the shape of the body. Examples include, but are not limited to fingerprint, face recognition, DNA, hand and palm geometry and iris recognition and behavioral which is related to the behavior of a person. Examples include, but are not limited to typing rhythm, gait, and speech pattern. Some researchers have coined the term behaviometrics for behavioral biometrics such as typing rhythm or mouse gestures where the analysis can be done continuously without interrupting or interfering with user activities.As a new technology of biometrics, gait recognition has recently gained more and more interests from researchers. Gait recognition is a process in which human motion features are extracted automatically and applied to recognize passerby's identity. It offers the possibility to identity people at a distance without any interaction or co-operation from the subject; this is the property which makes it attractive as a method of identification.In this thesis, we'll base our gait recognition method using canonical view synthesis (CVS) and gait energy image (GEI). Since dependency to the direction of walking is one of the gait recognition challenging problems, techniques such as canonical view synthesis will be used to better cope with view angle change. CVS is specifically used to reduce dependency to the direction of gait features through transformation using planar homography. GEI is then selected for gait representation, which is a spatio-temporal gait representation constructed using silhouettes. Finally in order to preserve the principal components, we use the Principal Component Analysis (PCA) to lower the dimensional feature space in which the characteristics of GEI are well-preserved.To identify individual, the outputs of the nearest neighbor classifiers are fused at the abstract level based on majority voting. The experimental results included, which is 100% for canonical view, demonstrate the encouraging performance of the proposed algorithm.