Header menu link for other important links
X
Matching digital and scanned face images with age variation
, , S. Bharadwaj, H.S. Bhatt, A. Noore
Published in
2010
Abstract
Existing face recognition systems have demonstrated success in constrained settings with limited variability in illumination, pose, and expression. However, these incremental improvements are not sufficient to transcend the challenging applications such as identifying missing persons or matching individuals with photo ID. These applications require recognition of face images with aging variations and matching digital to scanned photo images. This paper presents a preprocessing framework to enhance the quality of the input scanned and digital face images and minimize the aging differences. Three face recognition algorithms are used to evaluate the efficacy of the proposed framework. Experimental results computed on a digital and scanned database of 310 subjects show that the framework improves the accuracy of all three algorithms by minimizing the quality differences and the variations due to aging. © 2010 IEEE.
CONTINUE READING
Figures & Tables (8)
  • Figure-0
    Fig. 1. Digital and scanned face images with image quality ... Expand
  • Figure-1
    Fig. 2. Block diagram illustrating the steps involved ... Expand
  • Figure-2
    Fig. 4. Scanned face image and transformed digital face image
  • Figure-3
    Fig. 5. Results of the age transformation algorithm.
  • Figure-4
    Fig. 3. Examples of enhanced scanned and digital face images ... Expand
  • Figure-5
    TABLE I DETAILS OF THE FACE DATABASE USED FOR VALIDATION
  • Figure-6
    Fig. 6. ROC plot showing the performance of the proposed ... Expand
  • Figure-7
    TABLE II VERIFICATION PERFORMANCE OF THE THREE FACE VERIFICATION
About the journal
JournalIEEE 4th International Conference on Biometrics: Theory, Applications and Systems, BTAS 2010