A comparison of eigenface and HOG feature extraction for gender classification of facial images | |
| Author | Siriporn Chinsawat |
| Call Number | AIT Thesis no.CS-15-04 |
| Subject(s) | Genders--Classification Human face recognition (Computer science) |
| Note | The degree of Master of Science in Computer Science, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. CS-15-04 |
| Abstract | In recently Gender classification based facial image is play as an importance role in biometric identification. Gender classification is a process that identification by facial image. In real time face recognition have large database so gender classification can be used as indexing to reduce search space and time of processing by divide population into two halve based on gender. In this thesis we propose gender classification system which applied Eigenface and Histogram Oriented Gradient method. We compared performance of two feature extraction algorithms to determine the more efficient algorithm of gender classification from face image. For HOG method experiment by vary cell size to 2x2 and 4x4. Face database FERET have perform for experimental which be normalized. The images training set obtained 794 face image consist of 486 male face images and 308 female face images for each criteria. The test set obtained combine of male and female face images total 200 images. Support Vector Machine (SVM) use as classifier for classification process. The system applied by MATLAB 2015a version. Experimental result based on FERET database show HOG method is more efficient approach for gender classification than Eigenface. The HOG perform accuracy rate 87% with cell size 2x2 and 83.5% accuracy rate on Eigenface method. |
| Year | 2015 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis : no. CS-15-04 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Mongkol Ekpanyapong; |
| Examination Committee(s) | Dailey, Matthew N.;Chumpol Krootkaew; |
| Scholarship Donor(s) | Asian Institute of Technology Fellowship;Western Digital (Thailand) Co., Ltd.; |
| Degree | Thesis (M. Sc.) - Asian Institute of Technology, 2015 |