Dissertation > Excellent graduate degree dissertation topics show

Improved LDB Method of Face Recognition Based on Moments Property and Its Application

Author: ZhouLei
Tutor: WuLingDa
School: National University of Defense Science and Technology
Course: Control Science and Engineering
Keywords: face recognition feature extraction wavelet packet moment property LDB
CLC: TP391.41
Type: Master's thesis
Year: 2007
Downloads: 66
Quote: 0
Read: Download Dissertation

Abstract


Face recognition is an important embranchment problem in the fields of pattern recognition, image processing and artificial intelligence. It has been widely applied and highly valued by academic field, enterprise field, government and the military day by day. Generally, face recognition contains three steps: face image preprocessing, feature extraction and classifer design. Face feature extraction is a key step for face recognition. The goal is to represent high dimension face patterns in low dimension feature space so as to extract face feature with discriminant power for classification. Effective face representation methods will not only to design a better classifier, but also to improve recognition rate. However, since face patterns are complicated and multiform, the within-class scatters of face images are larger than the between-class scatters under various conditions, such as changing illumination, pose and facial expression, resulting that it is very difficult to represent face effectively. Therefore, face representation is also as the primary difficulty in face recognition.In this paper, the traditional LDB method is improved to be a new face recognition method, which is improved LDB method of face recognition based on moment property. The major contributions of this thesis are summarized as follows.1. Improve the definition of separability. The traditional LDB chooses the parts of coefficients after the face image decomposition with Wavelet packet as eigenvector, but this criterion computation does not satisfy with the separability’s linear additive property, this paper improves the definition of separability, it is called difference separability, which is used the average between-class distance difference with the coefficients correspondence average within-class distance, as the characteristic coefficient to certain the subbands of Wavelet packet decomposition.2. Improve the recognition effect by choosing the subbands. Because of not all subbands of wavelet and all decomposition coefficients are helpful for recognition, only the first and second origin moments of chosen coefficients in chosen subbands are used as the feature of faces;3. Improve the classification method. In the classification recognition method, this paper proposes a new classification method with the distance of weighted subband classification eigenvector, to substitute the nearest neighbor classification in the LDB.4. Using the method proposed in the ethesis, the module of face detection and recognization are designed and carried out which is an important composing of the digital media information processing system.There are two targets in face recognition: enhance recognition correctness and lower training and recognition time. In this paper, the traditional LDB method is improved to be a new face recognition method. This new method reduces the calculation complexity, lowers the training and recognition time and achieves real time recognition, while the useful face features are better described for classification and recognition correctness are enchanced.The experiments show that the proposed method in this paper has better recognition effect, the training and the recognition time separately reduced two magnitudes compared to the LDB method.

Related Dissertations

  1. Research on Automatic Detection Algorithm for Substructure Distress of Highway Pavement Based on SVM,U418.6
  2. Research on Algorithms of 2D Face Template Protection,TP391.41
  3. Research on Transductive Support Vector Machine and Its Application in Image Retrieval,TP391.41
  4. Research on Feature Extraction and Classification of Tongue Shape and Tooth-Marked Tongue in TCM Tongue Diagnosis,TP391.41
  5. Research on Feature Extraction and Classification of Doppler Ultrasonic Blood Flow Signals of Wrist Radial Artery,TP391.41
  6. An Approach for Identifying a Plant Resistance Gene Based on the Random Forest,Q943
  7. Tobacco Diseases Auto-Recognition Research Based on Image Processing Technology,S435.72
  8. Research on Nondestructive Detection Technology for External Qualities of Papayas Based-on Vision,S667.9
  9. Research for Infrared Image Target Identification and Tracking Technology,TP391.41
  10. Research of Diagnosing Cucumber Diseases Based on Hyperspectral Imaging,S436.421
  11. Research of Video Face Recognition Based on Weighted Voting and Key-Frame Extraction,TP391.41
  12. Face Recognition Method Based on DE,TP391.41
  13. The Speech Enhancement System Based on Binary Mask and Perceptual Wavelet Packet Transform,TN912.35
  14. Based on Sub-pattern Locality Preserving Projection Approaches Research for Race Recognition,TP391.41
  15. Research of Sparse Represention Based Face Recognition Algorithm,TP391.41
  16. A Novel Approach for Projection by Optimize Local Area in Face Recognition Applications,TP391.41
  17. Research on Deep Structure Learning Algorithms Based on Dynamic Fuzzy Relation,TP181
  18. Research of Fault Diagnosis Method of Analog Circuit Based on Improved Support Vector Machines,TN710
  19. The Study of 3d Face Recognition System,TP391.41
  20. 3D Face Recognition Based on Biomimetic Pattern Recognition,TP391.41
  21. Based on Lifting Scheme Wavelet Packet Transform and Artificial Neural Network Wing-box Multi-damage Detection,V224

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