Modified Illumination Invariant Algorithm Based Human Face Detection

Kaushik, M K and Samanth, Rashmi and Nayak, Subramanya G (2020) Modified Illumination Invariant Algorithm Based Human Face Detection. International Journal of Scientific & Technology Research, 9 (4). p. 592. ISSN 22778616

[img] PDF
8678.pdf - Published Version
Restricted to Registered users only

Download (188kB) | Request a copy

Abstract

Face recognition systems are adversely affected by multiple external factors like illumination, pose, quality of image etc. Illumination is one of the main factors that hinders the performance of the system, the proposed methodology minimizes the effects of varied lighting conditions on the images. The algorithm uses Histogram of Oriented Gradients (HOG) technique for feature extraction and Error correcting output Codes (ECOC) Multi class Support Vector Machines (SVM) for classification. The results are verified using ORL and Extended Yale Face Database B image sets and good accuracy was obtained in the classification of poor and well illuminated images.

Item Type: Article
Uncontrolled Keywords: Face recognition, Illumination, Histogram of Oriented Gradients, Error Correcting Output Codes, Support Vector Machines, ORL Database, Extended Yale Face Database
Subjects: Engineering > MIT Manipal > Electronics and Communication
Depositing User: MIT Library
Date Deposited: 16 Jul 2020 09:42
Last Modified: 16 Jul 2020 09:42
URI: http://eprints.manipal.edu/id/eprint/155438

Actions (login required)

View Item View Item