Fusion of Local and Global Features for Gender Classification
| Autor: | Chih-Wei Liu, 劉致緯 |
|---|---|
| Rok vydání: | 2013 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 101 In recent years, the technology of face recognition has been attracting an encouraging deal of research. In addition, the gender recognition also gains a lot of attention, and is widely used in business and surveillance fields. This paper proposes using facial features for gender classification. The gender classification is divided into two stages. Before the first stage, the pretreatment will complete the normalization of the input image. During the first stage, it will search unique features and determine the male. The second stage will use PBS (Path Binary Sequence) and LBP (Local Binary Patterns) to extract facial features. The extracted features are input into SVM (Support Vector Machine). SVM classifies the input data and decides the gender information. Our experiments use the FEI and FERET databases. The test method is the five-fold cross-validation. FEI and FERET database can yield gender recognition rate 92.5% and 91.0%, respectively. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|