Research On Local Linear Discriminant Analysis And Face Recognition Application Projection

In recent years, face recognition technology has been payed more attention. Face recognition usually consists of the following modules, namely face data acquisition, preprocessing, feature extraction and face recognition, and the feature extraction is a very important part, it has a great impact on the identification work. Over the past few decades, scholars have raised a number of related algorithms. These algorithms can be summarized into two categories, one is based on the reconstruction of feature extraction, in which a representative of algorithm is Principal Component Analysis; the other is based on the identification of feature exaction, in which a representative of algorithm is Fisher Linear Discriminant. Although these algorithms have a wide range of applications, but they still exist many problems, such as lack of robustness to external disturbances.Secondly, computer hardware are updating very fast, face recognition can be implemented on android mobile platform.Based on the achievements of predecessors, we propose some improved algorithms for their shortcomings, at the same time, we analysis and design a APP on android platform. The main work for this thesis can be summarized as follows:1?Local Linear Discriminant Analysis Using ?2-GraphA recently proposed method, called Local Fisher Linear Discriminant Analysis (LLDA), the experiment showed that compared with the traditional Fisher Linear Discriminant Analysis, it has a better result. However, it uses Euclidean distance selecting nearest neighbor samples which has some flaws in the way, such as the robustness is not good and not sparse, and so on. The paper presents an improved approach, called Local Linear Discriminant Analysis using ?2-Graph(?2 G_LLDA). It remains reconstructed coefficient of samples to select the nearest samples, which enhances the robustness of the algorithms and makes it sparse. The extensive experimental results in the ORL face database, YALE face database and AR face database have demonstrated the effectiveness of the proposed algorithm.2?VLLDA Based On Kernel Method For Face RecognitionIn this paper, a face recognition method by combing KPCA with VLLDA is proposed. Firstly, each train image in training set is mapped by using KPCA. Then, exception manifolds of training set are obtained by VLLDA. Finally, get the image under test set belongs to the probability of each classification, so as to identify its face categories. In the AR face database and ORL face database on the experimental results show that this method is more simple and effective and good recognition effect.3?Face Recognition Method By Combing 2DPCA With CRC_RLSa face recognition method by combing 2DPCA with CRC?_RLS is proposed. Firstly, each train image in training set is mapped by using 2DPCA. After projection, the training sample will be pulled into a column vector. Then, using CRC_RLS algorithm, we will get the sparse representation coefficients and reconstruction error. Finally, get the image under test set belongs to the probability of each classification, so as to identify its face categories. In the ORL face database and YALE face database on the experimental results show that this method is more simple and effective and good recognition effect.4?Android Smart Face Recognition System Analysis and DesignWith the development of the android platform, our life are changing, such as we use the mobile phones for transactions with Alipay, which means that demand for mobile phone-based privacy has a huge potential market. Meanwhile, the hardware of android smart phones are upgrading very fast, so that the phone has a faster calculation speed, higher resolution camera and more memory, which provides the conditions for android-based face recognition applications. The system is based on android platform, using face recognition technology to provide users with a complete set of services to help users manage photos, to protect users'privacy. This system include:1) Create a local face database, mainly through the camera of a mobile phone for photo acquisition, and then saved to the SD-Card.2) Face detection, including single and multi-detection test.3) Recognition, through the front camera to capture the face, determine whether it is registered, if he is a registered user, give his information.