Dissertation > Excellent graduate degree dissertation topics show

Study on Pose-varied Face Recognition Based on Subspace

Author: HuangZuoYing
Tutor: GongWeiGuo
School: Chongqing University
Course: Instrument Science and Technology
Keywords: Multi - pose face recognition Face posture correction Virtual samples Subspace methods
CLC: TP391.41
Type: Master's thesis
Year: 2009
Downloads: 151
Quote: 0
Read: Download Dissertation

Abstract


The face recognition technology is the use of computer analysis of face images, and then extract the effective characterization of face image recognition information is used to identify the person's identity, in order to achieve a technical supervision, management and control objectives, is a biometric technology a research focus of the current field of pattern recognition and image processing. Initiative, non-invasive and user-friendliness, etc. Compared with other biometrics, face recognition and attention by the researchers, can be widely used in the field of commercial, judicial, and public utilities. Therefore, in-depth study of the technology is important. The face recognition technology is often encountered in samples of high dimensionality, and the number of categories, less training samples, as well as pose, illumination, affects expression. Numerous methods of feature extraction, subspace method has a small computational cost, description ability separability, has now become the mainstream method of face recognition. In this paper, the test sample attitude change of face recognition, and can only get a small amount of training samples focus subspace feature extraction method by means of multi-pose face recognition research. Main research work as follows: ① In practical applications, the face recognition system by the posture change the impact of test face images, and the recognition accuracy to reduce this difficulty, studied based on the modification of the sine transform (Sine Transform, ST). The posture correction Face Recognition strategy. The method is the traditional face recognition system posture correction of this front-end processing, multi-gesture sample correction for frontal face image. This method can retain the texture information of the face image, to achieve fast face posture correction, and the computational cost, and easy to implement. This is in the case of only a small amount of positive training samples using the mainstream algorithm and also to ensure the robustness of the system. Compared with traditional face recognition system, by the interference of posture change and increase the posture correction module and will not increase system load, and also possible to improve the recognition accuracy. (2) In some occasions each person can only provide a positive training samples, but most of the existing face recognition algorithms to identify the level of rates is positively correlated with the number of training samples of each person. In this paper, polynomial transform and sine transform method to increase the virtual sample recognition rate is increased by increasing the number of training samples. Adding virtual samples to solve the problem of within-class scatter matrix zero, making various methods based on linear criterion can also use the single training sample face recognition problem. ③ subspace based feature extraction methods and under different conditions pose face recognition strategy combining verify the recognition rate of several sub-space algorithm face recognition strategy in posture correction, and the recognition results before using posture correction compared to the recognition rate has been improved.

Related Dissertations

  1. The Study on Face Recognition Technology Based on Subspace Methods,TP391.41
  2. Research on Content Based Handmetric Identification,TP391.41
  3. Research of Intelligent Digital Monitor System Based on Face Detection and Recognition,TP277
  4. The Study of Parallel Algorithms for the Solution of Special Structured Large Linear System of Equations under Distributed Memory Environment,TP301.6
  5. Numerical Analysis of Special Matrices and Preconditioning Techniques for Iteration Solutions of Saddle Point Problems,O151.21
  6. Palmprint Recognition Based on Subspace Methods,TP391.41
  7. Research of Facial Landmark Location and Face Recognition,TP391.41
  8. Solving some special numerical solution of sparse linear systems,O241.6
  9. The Researches of Iterative Methods and Preconditioning Techniques for Linear Systems and Saddle Point Problems,O241.6
  10. Subspace Analysis and Classification Method for Face Recognition,TP391.41
  11. Research on Face Recognition Using Subspace Methods,TP391.41
  12. The Research of Numerical Solution for Large Sparse Algebraic Systems,O241.6
  13. Preconditioning Method and Its Applications in Numerical Simulation for Electromagnetic,O441.4
  14. Iterative Methods for Linear Systems and Preconditioning Techniques with Applications in Electromagnetic Scattering Computing,O441.4
  15. Multi-Feature Analysis of Face Detection and Recognition and System Implementation,TP391.41
  16. Recognition based on the time-varying characteristics and nuclear methods for radar target distance like,TN957.5
  17. Restarted Simpler GMRES Augmented with Approximate Errors,O241.6
  18. Research on Channel Estimation and Equalization Methods in MIMO-OFDM System,TN919.72
  19. Bi-iteration Subspace Tracking,TN953
  20. Study on Preconditioning Techniques for Generalized Minimal Residual Algorithm,O241.6

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile