Face Recognition Based on Gender Using a Modified Method of 2D-Linear Discriminant Analysis

摘要

Currently, the development of technology related to facial recognition is already widely used in applications introduction of biological data (biometrics) such as the introduction of gender. The introduction of a gender-based human face image is part of a biometric method for identifyingan individual by taking facial features. The implementation of applications that require the introduction of gender is the process of market segmentation to determine the demographic trend of products marketed by sex. It also can be used to limit access to a room and others. The study aimsto develop a system which can be used to recognize gender-based on facial image. Through this system, the recognizing process of gender can be identified accurately and quickly. This research used to test data 200×200 pixels. The data contains 400 facial images which consist of 200 maleimages and 200 female images. Two-Dimensional Linear Discriminant Analysis Modification (Modification 2DLDA) method was used for feature extraction. This method is resulted from the development Two-Dimensional Linear Discriminant Analysis (2DLDA) method, in which it comes from the developmentof Linear Discriminant Analysis (LDA) method. Also Two-Dimensional Correlation Coefficient method used to classify gender recognition by calculating the value of a strong existence or inexistence of linear relationship between two variables. As a result, the highest accuracy of gender recognitionis 91,45% while the lowest accuracy rate is 86,33%.