Face images are influenced by illumination, pose, age changes and other effects, and it is difficult u-sing a single feature extraction method to obtain high accuracy of sex recognition. In order to improve the correct rate of gender recognition, the geometric characteristics and principal component analysis were combined with the gender recognition algorithm. First, geometric features method was used for face image feature extraction. Then the principal component analyses was use to select the characteristics which have important implications for the identification re-sults. Finally, the characteristics were input to the support vector machine for learning and the gender classifier was established. The Indians face base was used with the algorithm for performance tests. The results show that this algo-rithm can accelerate the gender recognition speed and improve the correct rate of recognition under larger chang in il-lumination and pose.%研究性别识别问题,人脸图像受到光照、姿态、年龄的变化等影响,采用单一特征提取方法难获得较高的性别正确率.为提高性别识别正确率,提出采用几何特征和主成分分析结合的性别识别算法.首先采用几何特征方法对人脸图像的特征进行提取,然后采用主成分分析选择对识别结果有重要影响的特征,最后将选择特征输入到支持向量机进行学习,建立性别分类器.采用印度人脸库对算法性能进行检验,结果表明,本文算法加快了性别识别的速度,提高了识别正确率,能够对光照和姿态变化较大的图像进行正确识别.
展开▼
