You are here

Different Facial Recognition Techniques in Transform Domains

Download pdf | Full Screen View

Date Issued:
2018
Abstract/Description:
The human face is frequently used as the biometric signal presented to a machine for identificationpurposes. Several challenges are encountered while designing face identification systems.The challenges are either caused by the process of capturing the face image itself, or occur whileprocessing the face poses. Since the face image not only contains the face, this adds to the datadimensionality, and thus degrades the performance of the recognition system. Face Recognition(FR) has been a major signal processing topic of interest in the last few decades. Most commonapplications of the FR include, forensics, access authorization to facilities, or simply unlockingof a smart phone. The three factors governing the performance of a FR system are: the storagerequirements, the computational complexity, and the recognition accuracy. The typical FR systemconsists of the following main modules in each of the Training and Testing phases: Preprocessing,Feature Extraction, and Classification. The ORL, YALE, FERET, FEI, Cropped AR, and GeorgiaTech datasets are used to evaluate the performance of the proposed systems. The proposed systemsare categorized into Single-Transform and Two-Transform systems. In the first category, the featuresare extracted from a single domain, that of the Two-Dimensional Discrete Cosine Transform(2D DCT). In the latter category, the Two-Dimensional Discrete Wavelet Transform (2D DWT)coefficients are combined with those of the 2D DCT to form one feature vector. The feature vectorsare either used directly or further processed to obtain the persons' final models. The PrincipleComponent Analysis (PCA), the Sparse Representation, Vector Quantization (VQ) are employedas a second step in the Feature Extraction Module. Additionally, a technique is proposed in whichthe feature vector is composed of appropriately selected 2D DCT and 2D DWT coefficients basedon a residual minimization algorithm.
Title: Different Facial Recognition Techniques in Transform Domains.
0 views
0 downloads
Name(s): Al Obaidi, Taif, Author
Mikhael, Wasfy, Committee Chair
Atia, George, Committee CoChair
Jones, W Linwood, Committee Member
Myers, Brent, Committee Member
Moslehy, Faissal, Committee Member
University of Central Florida, Degree Grantor
Type of Resource: text
Date Issued: 2018
Publisher: University of Central Florida
Language(s): English
Abstract/Description: The human face is frequently used as the biometric signal presented to a machine for identificationpurposes. Several challenges are encountered while designing face identification systems.The challenges are either caused by the process of capturing the face image itself, or occur whileprocessing the face poses. Since the face image not only contains the face, this adds to the datadimensionality, and thus degrades the performance of the recognition system. Face Recognition(FR) has been a major signal processing topic of interest in the last few decades. Most commonapplications of the FR include, forensics, access authorization to facilities, or simply unlockingof a smart phone. The three factors governing the performance of a FR system are: the storagerequirements, the computational complexity, and the recognition accuracy. The typical FR systemconsists of the following main modules in each of the Training and Testing phases: Preprocessing,Feature Extraction, and Classification. The ORL, YALE, FERET, FEI, Cropped AR, and GeorgiaTech datasets are used to evaluate the performance of the proposed systems. The proposed systemsare categorized into Single-Transform and Two-Transform systems. In the first category, the featuresare extracted from a single domain, that of the Two-Dimensional Discrete Cosine Transform(2D DCT). In the latter category, the Two-Dimensional Discrete Wavelet Transform (2D DWT)coefficients are combined with those of the 2D DCT to form one feature vector. The feature vectorsare either used directly or further processed to obtain the persons' final models. The PrincipleComponent Analysis (PCA), the Sparse Representation, Vector Quantization (VQ) are employedas a second step in the Feature Extraction Module. Additionally, a technique is proposed in whichthe feature vector is composed of appropriately selected 2D DCT and 2D DWT coefficients basedon a residual minimization algorithm.
Identifier: CFE0007146 (IID), ucf:52295 (fedora)
Note(s): 2018-08-01
Ph.D.
Engineering and Computer Science, Electrical Engineering and Computer Engineering
Doctoral
This record was generated from author submitted information.
Subject(s): Face Recognition -- Transform Domain -- DWT -- DCT
Persistent Link to This Record: http://purl.flvc.org/ucf/fd/CFE0007146
Restrictions on Access: public 2018-08-15
Host Institution: UCF

In Collections