Fyp Proposal
Project title * Pose-Invariant Face Recognition
Abstract * Project motivation: To recognize pose-invariant face images. * Objectives: Research and implement algorithms to get a better recognition performance of variant poses.
Background * Face recognition is very useful in some area. But even the same person would have variety of poses. Pose has become an important factor affecting face recognition. The key point is to get the face feature which is invariant along with the changing poses.
Methodology * Software: Visual Studio 2010 and MATLAB 2010
Face Recognition Methods * Principle Component Analysis (PCA)
PCA is a method which can reduce the large dimensionality of the data. It can find several unrelated basis vectors to replace the large amount of variables before and the new variables can also represent the primary information as faithfully as possible. * Multiview Face Hallucination (MFH)
MFH is a simple and efficient method to generate high-resolution multiview faces from a single low-resolution one. * Linear Discriminant Analysis (LDA)
LDA is a common method used in face recognition. The principle is to optimize the Fisher discriminant function, which maximize the ratio between between-class scatter and within-class scatter of samples. * Locality Preserving Projection (LPP)
LPP is an algorithm for learning a locality preserving subspace that is to preserve the local structure of the image space. * Canonical Correlation Analysis (CAA)
CAA is a method that makes good use of the correlation between the basis vectors to reflect the overall correlation between two sets of data. In order to find the correlation between two sets of variables, several representative basis vectors can be extracted and the overall correlation can be represented.
Implementation Plan * Semester 1 * PCA (done) * Simple multiview algorithm (November 2012 – December 2012) * Eigen-transformation (November 2012 –
Abstract * Project motivation: To recognize pose-invariant face images. * Objectives: Research and implement algorithms to get a better recognition performance of variant poses.
Background * Face recognition is very useful in some area. But even the same person would have variety of poses. Pose has become an important factor affecting face recognition. The key point is to get the face feature which is invariant along with the changing poses.
Methodology * Software: Visual Studio 2010 and MATLAB 2010
Face Recognition Methods * Principle Component Analysis (PCA)
PCA is a method which can reduce the large dimensionality of the data. It can find several unrelated basis vectors to replace the large amount of variables before and the new variables can also represent the primary information as faithfully as possible. * Multiview Face Hallucination (MFH)
MFH is a simple and efficient method to generate high-resolution multiview faces from a single low-resolution one. * Linear Discriminant Analysis (LDA)
LDA is a common method used in face recognition. The principle is to optimize the Fisher discriminant function, which maximize the ratio between between-class scatter and within-class scatter of samples. * Locality Preserving Projection (LPP)
LPP is an algorithm for learning a locality preserving subspace that is to preserve the local structure of the image space. * Canonical Correlation Analysis (CAA)
CAA is a method that makes good use of the correlation between the basis vectors to reflect the overall correlation between two sets of data. In order to find the correlation between two sets of variables, several representative basis vectors can be extracted and the overall correlation can be represented.
Implementation Plan * Semester 1 * PCA (done) * Simple multiview algorithm (November 2012 – December 2012) * Eigen-transformation (November 2012 –
References: [1] D.H. Liu, L.S. Shen & K.M. Lam, "Face Recognition：A Survey," J. Circuits Syst., vol. 9, no. 1, pp. 86–94, Feb. 2004 [2] Kim K (2003), Face recognition using principle component analysis. Department of Computer Science, University of Maryland, College Park [3] X. Ma, H. Huang, S.P. Wang & C. Qi, "A Simple Approach To Multiview Face Hallucination," Ieee Signal Processing Lett., vol. 17, no. 6, pp. 579–582, Jun. 2010 [4] D. Weenink, Canonical Correlation Analysis, Institute of Phonetic Sciences, University of Amsterdam, Proceedings 25 (2003), 81–99