Face Pose and illumination Normalization for Unconstraint Face Recognition from Direct Interview Videos
T. Shreekumar1, K. Karunakara2

1T. Shreekumar, Department of Computer Science and Engineering, Mangalore Institute of Technology & Engineering, Mangalore (Karnataka), India.
2K. Karunakara, Department of Information Science and Engineering, Sri Siddhartha Institute of Technology, Tumkur (Karnataka), India.
Manuscript received on 22 April 2019 | Revised Manuscript received on 04 May 2019 | Manuscript Published on 17 May 2019 | PP: 59-66 | Volume-7 Issue-6S4 April 2019 | Retrieval Number: F10120476S419/2019©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. 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: The Posture variation and the Lighting sets in the vicinity problem are the two major challenges in Facial parameters identification. We use Local Linier Regression (L.L.R) and Discrete Cosine Transform for Posture and Lighting set correction. In this publication Principal Composition Exploration (P.C.A), Fisher’s Linier Discrimination examination (F.L.D.A) and the combined score of P.C.A and L.D.A are castoff to articulate the single layered sped forward Neural Network. During the testing portion Posture and the Lighting sets in the vicinity normalization are carried out using L.L.R and D.C.T respectively. Then the combined score of P.C.A and L.D.A of test image is used to recognize the image using the trained Neural Network. We further use the Support Vector Machine (S.V.M) to train and recognize the Face images replacing Neural Network. We are able to obtain more improvement in the results such as computational complexities and computation speed.
Keywords: Artificial Neural Network, Support Vector Machine, Discrete Cosine Transform, Fisher Linear Discriminant Examination, Linear Facial parameters identification, Principal Compound Investigation, Local Linear Regression.
Scope of the Article: Pattern Recognition