Manuscript Number : IJSRSET173886
Face Recognition using Features from Accelerated Segment Test for Invariance Towards Changes in Illumination
Authors(1) :-Aruna Bhat
The area of face recognition in spite of being the most unobtrusive biometric authentication methodology has enjoyed limited application is real world owing to constant changes that occur on the face due to various factors. The changes brought about on a face image due to variations in the illumination have been known to be a major contributor towards tremendous changes in the image which render any face recognition system of little practical use. This paper presents a new technique to address this issue by identifying robust illumination invariant fiducial points from the face so that changes in the parameters of illumination do not impact the crucial information in a face image and thus make the face recognition system stronger and usable in practice. We have made the use of Features from Accelerated Segment Test (FAST) for detecting the interest points in the face images from the training and the test data sets. It identifies those fiducial points in a face image which are located at well-defined positions irrespective of changes happening in the image due to variations in illumination. The technique is also computationally far more efficient than various other standard methods that have been used so far for feature extraction. Using nearest neighbour search strategy, each fiducial point in the test feature vector is compared with the fiducial points in every training feature vector. The maximum number of matches between the test and the training feature vectors leads to a possible match. FAR and FRR for the method were observed to showcase encouraging results in authenticating faces even in presence of extreme variations in illumination conditions which supports the use of FAST features for designing face recognition methodologies that are significantly invariant to the changes in illumination.
Aruna Bhat
Fiducial points, Features from Accelerated Segment Test, NNS, FAR, FRR, DCT, Homomorphic filtering
Publication Details
Published in :
Volume 3 | Issue 8 | November-December 2017 Article Preview
Research Scholar, Department of Electrical Engineering, IIT Delhi, India
Date of Publication :
2017-12-31
License: This work is licensed under a Creative Commons Attribution 4.0 International License.
Page(s) :
845-851
Manuscript Number :
IJSRSET173886
Publisher : Technoscience Academy
Journal URL :
https://ijsrset.com/IJSRSET173886