1 / 25

Preprocessing to enhance recognition performance in the presence of: Illumination variations Pose /expression/scale va

Preprocessing Overview. Preprocessing to enhance recognition performance in the presence of: Illumination variations Pose /expression/scale variations - Resolution enhancement (deblurring) Stand-alone recognition system Preprocessing/recognition results

ashton
Download Presentation

Preprocessing to enhance recognition performance in the presence of: Illumination variations Pose /expression/scale va

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. Preprocessing Overview • Preprocessing to enhance recognition performance in the presence of: • Illumination variations • Pose/expression/scale variations - Resolutionenhancement (deblurring) • Stand-alone recognition system • Preprocessing/recognition results - Face Recognition Grand Challenge (NIST)

  2. Face Recognition Problem: Current state-of-the-art face recognition systems degrade significantly in performance due to variations in pose, illumination, and blurring. Solution: PREPROCESSING RESTORATION/ ENHANCEMENT FACE RECOGNITION SYSTEM IMAGE CAPTURE POSE CORRECTION due to mismatch in facial position, facial expression and scale ILLUMINATION CORRECTIONdue to mismatch in lighting conditions in both indoor and outdoor environments DEBLURRINGdue to mismatch in camera focus, camera lenses, camera resolution and motion blur

  3. Highlights of the approach • No a priori information with regards to pose orientation, camera parameters, etc • No laser scanned images for 3D reconstruction • No manual detection of feature points • Preprocessing & Stand-Alone Recognition

  4. Principle • Find a function which maps a given test (probe) image into the correct train (gallery) image • Approach where M is the number of training images • Select that is maximally bijective

  5. domain (X) f range(X) Y X One to One and Onto (bijection) Recognition Principle • A function ’f ‘ is found which maps points in the test (probe) to equivalent points in the train (gallery) where X = Test image (domain) Y = Train image (co-domain) = Bijective function mapping X  Y Test Train

  6. where Y = Train image (domain) X = Test image (co-domain) = Bijective function mapping Y  X Inverse Estimation • A function ’g ‘ is found which maps points in the train (gallery) to equivalent points in the test (probe) domain (Y)  g range(Y) Y X Test Train

  7. Blue,Green,Cyan  Red  Measure of Bijectivity X f Y Partition X where n is the total number of distinct blocks in X

  8. g X Y Blue,Green,Cyan  Red  Partition Y Measure of Bijectivity where p is the total number of distinct blocks in Y

  9. = Forward (test  train) = Backward (train  test) The Bijectivity score is given by: = Adaptive Forward (test  train) = Adaptive Backward (train test) Measure of Bijectivity = constants and

  10. Mapping Properties Train Image No.1

  11. Preprocessing Example

  12. Preprocessing Performance

  13. Face recognition performance Table 4.2 Yale Results-Stand Alone Recognition-II

  14. Face recognition performance Table 4.4 PIE Database-Stand Alone Recognition PCA-based Approach recognition accuracy: 5.88 %

  15. Preprocessing for Illumination Correction Training Test Preprocessed

  16. Preprocessing for Illumination Correction • Algorithm based on image adaptive least squares illumination correction Training image A Testing image B Adaptive segmentation Least squares estimate of illumination Image A illuminated as B Image B illuminated as A

  17. Preprocessing ResultsIllumination Correction Enrollment : Process of accepting the image and creating a feature set for recognition.

  18. Comparison with Existing Methods • 3D morphable models : • Good results (FRVT 2002). • Very complex, computationally expensive, • manual labeling of features 1. T. Vetter and V. Blanz, “Face Recognition Based on Fitting a 3D Morphable Model,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 25, no. 9, pp. 1063--1075, Sept. 2003.

  19. train test Vector Field Representation of f Preprocessed Test Image Preprocessing Example

  20. Preprocessing Example Notre Dame Database

  21. Preprocessing Example Notre Dame Database

  22. With the correct gallery Bijective Mapping Gallery Probe White Region  measure of bijectivity (52.91%) Recognition Example

  23. With the incorrect gallery Bijective Mapping Gallery Probe White Region  measure of bijectivity (33.94%) Recognition Example

  24. Face Recognition Performance Notre Dame Database

  25. Conclusion and Future Work • New algorithm for registration and illumination correction to enhance the performance of face recognition systems • Algorithm is based on properties of the mappingbetween test and train data • Mapping produces similarity scores which can be used for a stand-alone face recognition algorithm • Extend algorithm for high resolution data • Reduce algorithm complexity

More Related