Face Recognition Under Partial Occlusion.

Date of Submission

December 2009

Date of Award

Winter 12-12-2010

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Master's Dissertation

Degree Name

Master of Technology

Subject Name

Computer Science


Electronics and Communication Sciences Unit (ECSU-Kolkata)


Chanda, Bhabatosh (ECSU-Kolkata; ISI)

Abstract (Summary of the Work)

In today’s world identification of a person is very important and necessary mostly in security purpose. Now a days many organisation use different type of identification techniques, one of them is Face Recognition. But in different conditions they may not have the capability to recognise efficiently w.r.t both success rate and time complexity of their algorithm. There are different as well as important issues in face recognition like expression, ageing, occlusion that may make trouble to any recognition system. In this report we have presented a novel approach of recognizing human face from various type of partially occluded frontal views.The idea behind this approach is sparse representation that means the test face image should only be represented in terms of training face images of the same object. Here we consider that the occlusion is also sparse i.e. a fraction of the image pixels are occluded. We propose a simple and novel algorithm which uses pseudo inverse to express the test image as a sparse linear combination of training samples plus a sparse error due to occlusion. Lastly we use l2 norm and k-NN classifier to recognize the face. One advantage of this approach is no need of feature selection, dimension reduction, domain-specific information about the image. We analysis the experimental results of this algorithm which shows how much occlusion the algorithm can handle and how to choose the training data to maximize robustness for different type of occlusion. It also shows where the maximum information present in the face image. To verify the algorithm we use the publicly available ORL database and MATLAB tool in windows system.


ProQuest Collection ID: http://gateway.proquest.com/openurl?url_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&res_dat=xri:pqm&rft_dat=xri:pqdiss:28843237

Control Number


Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.



This document is currently not available here.