Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

Research about Access Control System Based on Face Recognition

Author XueXiaoLi
Tutor ZhuJinLing
School Southwest Jiaotong University
Course Pattern Recognition and Intelligent Systems
Keywords Face Detection Face Recognition Access Control System Adaboost Gabor SVM
CLC TP391.41
Type Master's thesis
Year 2011
Downloads 137
Quotes 1
Download Dissertation

In the past few decades, face detection and face recognition technology has made considerable progress, but still does not solve the core problem in the practical application of the process. The reason is that (1) face detection algorithm itself is relatively time-consuming, which makes the whole face recognition system can not meet the practical application of real-time requirements; (2) of the existing face recognition algorithm generally has high dimensional training set and the sample nature; (3) facial feature class is sometimes greater than the class change; (4) lack of efficient facial feature description and the corresponding high-precision core recognition algorithm; (5) how to improve the face recognition system inevitable mis-registration error robustness; addition, the practical application process camera image quality, the efficiency of algorithms, ambient light dramatic changes are all problems that can not be ignored. In this paper, we discuss in detail the problem and the solution. Firstly, face detection, face recognition developments in the field of history and the current development in recent years made a brief introduction. Next, the article describes the face detection algorithm based on haar features and AdaBoost cascade classifier, and the amount of computation of the algorithm, operating efficiency analysis. Match the two existing face detection, motion detection and edge direction to improve, reducing the computation and improve the operating efficiency of the algorithm. Then, the article more popular in recent years Gabor feature extraction methods were described in detail, and proposed to the Gabol-Sobel and LTP method of combining the facial feature extraction. By way of comparison, the improved method compared to the original Gabor feature extraction method with class aggregation class dispersion characteristics, and thus more conducive to the final feature classification and discrimination. After that, the article elaborates on the limited sample the classical classifier-SVM learning classification. Resampling technique to improve the performance of the classifier boosting algorithm in-depth analysis, and on this basis, the proposed method of combining boosting resampling algorithm and SVM to improve the classification, at the same time reduce the interference of the sample nature, classification and discrimination ability. In addition, the article also put forward by the disturbance Facial positioning accuracy to generate multiple virtual face, thus increasing the training samples to improve the robustness of the system mis-alignment. Finally, the face detection, face recognition improved algorithm based on the design of a face recognition-based access control systems, with detailed descriptions of the system's organizational structure, the architecture of each sub-module, the system's advantages and disadvantages .

Related Dissertations
More Dissertations