A Gender Recognition System Using ASM Facial Feature Points

• Main Authors: Hsin-ru Lo, 羅心汝
• Other Authors: Chen-Chiung Hsieh
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/10530736583982562448

碩士 === 大同大學 === 資訊工程學系(所) === 104 === The application of gender recognition is in the filed of shopping center and public places. The recognition system aims to let advertisers know the proportion of male and female of the population when ads show displays. In this paper, Active Shape Model (ASM) is used to extract facial gender features for gender recognition. Traditional, there are neural network classifier, gender identify facial features, and Principal Components Analysis (PCA) classifier for gender identification. Here, we propose innovative facial and hair features for gender classification which could be divided into several steps. First, we use ASM to extract facial feature points. Second, analysis of gender differences on the human face is done for feature extraction. In gender identification, the main features are the beard, the hair position, the length of the hair, and the philtrum. Decision tree classification is used to determine the target is male or female. In the experiments, gender identification test using Psychological Image Collection at Stirling (PICS), FEI Face Database, and collected the face photos by ourselves, to deal with 7.65 photos per second, the gender identification accuracy is 97.0%, proved the practicability of the method.