Face Detection And Recognition System

As the society development, an accurate automatic personal identification is critical to a wide range of application domains such as access control, electronic commerce, and welfare benefits disbursement. Traditional personal identification methods (e.g., passwords, PIN) suffer from a number of drawbacks and are unable to satisfy the security requirement of our highly inter-connected information society. Biometrics refer to automatic identification technology of an individual based on their physiological traits such as fingerprint, face and iris or behavioral traits such as signature, speech and gait. Currently, there are many biometric techniques that are widely used. A biometric system is essentially a pattern recognition system, which makes a personal identification by establishing the authenticity of a specific physiological or behavioral characteristic of the user.Facial images are probably the most common biometric characteristic in order to make a personal identification. Face recognition is one of the most active research areas ranging from static, controlled mug shot verification to dynamic, uncontrolled face identification, in a cluttered background. Face recognition is a non-intrusive technique and people generally do not have any problem in accepting face as a biometric characteristic.Main contents of this thesis include three aspects:1 We present a method based on image gap measurement, target and background can be divided accurately by this method, also can divide many images overlap and different parts of a image can be divided accurately. Compared with traditional method, our method can save more time and improve division efficiency. At the same time, we conclude that main recent algorithm of face detection and recognition, and evaluation advantage and disadvantage of very algorithm.2 We present a face tracking method based on combination histogram matching with shape constrains. First we take face figure as ellipse, then an adaptive weighted histogram matching is used to estimate an initial position, in which an optimal method called mean shift is adopted to search matching path automatically. After histogram matching, a normalized gradient model of elliptical boundary is used to accurately track the head's position and scale size in a local range. Experiments demonstrate that it is a real-time and robust tracker.3 The most popular methods for face recognition are second-order methods, such as Principal Component Analysis (PCA), and Probabilistic PCA (PPCA). This means methods that find the representation using only information contained in the covariance matrix of the face data vector. The use of second-order techniques is to be understood in the context of the classical assumption of gaussianity. However, the distribution of face data vector in promising application must not be assumed to be gaussian. Higher-order methods use information on the distribution of face data vector is not contained in the covariance matrix. The conventional method based on higher-order statistics is independent component analysis in neural network. The classical application of Independent Component Analysis (ICA) model is blind source separation, because of the restriction of this ICA and of noise-free model, the parameters may be ill-posed in applications. In this thesis, based on the advantage of ICA and kernel technology, and we utilize nonlinear function of the Reproducing Kernel Hilbert Space, RKHS as contrast function, so we can map the low-dimensional space onto high-dimensional space, then we take use of kernel method to search the least value of contrast function in the space. Compared with the traditional ICA, the method of KICA will be more flexible and robust.