A Performance Study of SIFT, SIFT-PCA and SIFT-LDA for Face Recognition
Sanket Panda1, Shaurya Nigam2, Rohit Kumar3, Mamatha HR4
1Sanket Panda, PES Institute of Technology, 100 Feet Ring Road, BSK III Stage, Bangalore-560085, Karnataka, India
2Shaurya Nigam, PES Institute of Technology, 100 Feet Ring Road, BSK III Stage, Bangalore-560085, Karnataka, India
3Rohit Kumar, PES Institute of Technology, 100 Feet Ring Road, BSK III Stage, Bangalore-560085, Karnataka, India.
4Dr. Mamatha HR, Professor, Information Science Department, Protocol d’Echanges pour un Systeme Interbancaire de Telecompensation, Bangalore, India.
Manuscript received on June 19, 2015. | Revised Manuscript received on June 29, 2015. | Manuscript published on July 05, 2015. | PP: 66-72 | Volume-5 Issue-3, July 2015. | Retrieval Number: C2647075315
Open Access | Ethics and Policies | Cite
©The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/
Abstract: Humans possess an innate cognitive ability to recognize faces and identify persons in an instantaneous and effortless manner. Face Recognition (FR) aims to emulate this capability using automated mechanisms and has been at the crux of research efforts in the domain of computer vision for the past two decades. Even though this process of emulation is arduous, it holds considerable promise in terms of its potential applicability, and hence, FR has steadily received consistent mainstream attention. The human cognition system generally stores and recollects images instantly based on necessity and similarly, machine vision replicates this process by storing images in a database and accordingly requires to be competently trained in order to accurately recognize faces. In this regard, many diverse algorithms have been proposed over the years with varying effectiveness. Therefore in this paper, we meticulously compare the conventional SIFT features method with its Weighted PCA and LDA variants in order to investigate as to which approach is more potent.
Keywords: Eigenfaces; Fisherfaces; Face Recognition; SIFT; PCA; LDA.