Gender and ethnic classification of Malaysian face images for face recognition


Chan, Chee Suit (2011) Gender and ethnic classification of Malaysian face images for face recognition. Masters thesis, Multimedia University.

Full text not available from this repository.


With the advances in technology, there comes the need for increased security in various fields and applications. Out of the many physical attributes that can be found on every human being, a face and its features prove to be one of the most unique yet. The problem of face recognition can be affected by change in pose and illumination, but is also incrementally challenged when a database contains a large number of individuals. While recognition aims to identify a face image, classification on the same set of images attempts to identify the group to which a face image belongs. As such, while the classification of gender and ethnicity can provide for a demographic classification system on its own, this information can also be supplied to a face recognition system to narrow down its search. Hence, this work aimed to investigate and implement gender and ethnicity classification of face images and subsequently the application of these classifications to the recognition of face images. A database of Malaysian faces was constructed for the purpose of this study. It consists of 151 different individuals from both genders and three different ethnicities (Chinese, Indian and Malay). For each classification experiment, results were obtained from using varying degrees of the original number eigenvectors on all possible combinations between four feature extraction methods and seven classifiers.

Item Type: Thesis (Masters)
Additional Information: Call No.: TA1650 C43 2011
Uncontrolled Keywords: Human face recognition (Computer science)
Subjects: T Technology > TA Engineering (General). Civil engineering (General) > TA1501-1820 Applied optics. Photonics
Depositing User: Ms Nurul Iqtiani Ahmad
Date Deposited: 13 Jan 2016 10:57
Last Modified: 13 Jan 2016 10:57


Downloads per month over past year

View ItemEdit (login required)