1 / 22

PhD Thesis

PhD Thesis. Biometrics. Science studying measurements and statistics of biological data Most relevant application: id. recognition. Why Facial Biometrics ?. Most intuitive way of identification Socially and culturally accepted worldwide It may work without collaboration. 2001. 19.2 %.

hachi
Download Presentation

PhD Thesis

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. PhD Thesis

  2. Biometrics • Science studying measurements and statistics of biological data • Most relevant application: id. recognition

  3. Why Facial Biometrics ? • Most intuitive way of identification • Socially and culturally accepted worldwide • It may work without collaboration 2001 19.2 % 43.6 % 2006

  4. Facial Biometrics • Challenges ahead • Less accurate than iris and fingerprint • Problems with uncontrolled environments (illumination, viewpoint…) Best system Average Fully automatic

  5. Automatic training from examples User-defined template based on landmarks Model-based parametrization Generative models Active Shape Models T.F. Cootes, C. J, Taylor, D.H. Cooper, J. Graham (1995) Computer Vision and Image Understanding, 61(1):38–59

  6. This thesis… • Focus on 3 contributions to ASMs on relevant aspects for facial feature localization: • More accurate segmentation invariant to in-plane rotations • Add robustness to out-of-plane rotations • Estimate the Reliability of the segmentation 1 2 3 4

  7. Face outlines based on landmarks Shape statistics to learn spatial relations Texture statistics for image search Shape statistics Local texture statistics 1 ASM: Construction of the model PDM IIMs Landmarked Training Set

  8. 1.- The input shapes are aligned to remove scale, translation and rotation effects. 1 1 Point Distribution Model Model Coordinates Image Coordinates

  9. 2.- Principal Component Analysis (PCA) on the aligned shapes (2L)-space representation PCA-space representation 1 1 Point Distribution Model

  10. 1 1 Point Distribution Model (PDM) • Can determine valid shapes • Can get closest valid shape • Introduces a representation error

  11. More specific More general 1 1 Point Distribution Model (PDM)

  12. 1 1 PDM: Modes of variation Variation from 1st Principal Component

  13. 1 1 PDM: Modes of variation Variation from 2nd Principal Component

  14. First order derivatives of the pixel intensity For each landmark Sampled perpendicularly to the contour 1 1 ASM: Local Texture Statistics (1) i-th landmark

  15. Second order statistics for each landmark 1 1 ASM: Local Texture Statistics (2) i-th landmark

  16. The average shape is placed on the image, roughly matching the face position 1 1 ASM: Model Matching • Displacement of each landmark to minimize the Mahalanobis distance to the mean profil • Apply shape model restrictions

  17. 1 1 ASM: Model Matching Steps 2 and 3 are repeated a fixed number of iterations at different resolutions, increasing detail

  18. 1 1 ASM: Model Matching

  19. 1 1 ASM: Model Matching

  20. 1 1 ASM: Complex textures • Several factors modify facial appearance • beard, hair cut, glasses, teeth. • The distribution of the normalized gradient is often non Gaussian nor unimodal.

  21. 1 1 ASM: Complex textures

  22. Texture description based on Taylor series Grids centered at the landmarks for local analysis Non linear classifier (kNN) for inside-outside labeling outside inside 2 1 Optimal Features ASM B. van Ginneken, A.F. Frangi, J.J. Staal, B.M. terHaarRomeny, and M.A. Viergever (2002) IEEE Transactions on Medical Imaging, 21(8):924–933

More Related