融入性别信息的人脸识别方法
首发时间:2012-11-20
摘要:人脸识别作为一种典型的生物特征识别技术,近年来得到了长足的发展。但是,对于绝大多数人脸识别方法而言,都存在一个通病,那就是存在性别上的误判,即将男性(女性)识别成女性(男性)。因此,为了提高人脸识别的准确率,本文在人脸识别算法的基础上,融入性别信息进行识别。融入性别信息的人脸识别算法就是:在原有算法的最后分类识别阶段,对输入人脸图像进行性别识别后,仅在人脸数据库中与输入人脸图像同性别的样本中进行分类识别。实验结果证明:在融入性别信息的基础上,人脸识别算法的准确率得到了极大的提高。
For information in English, please click here
Face recognition by fusing gender information
Abstract:As a typical biometric technique, face recognition has developed fast in recent years. However, the problem of gender misjudgment is common to most face recognition methods. That is to say, it often recognizes a man (woman) as a woman (man). In order to improve the recognition rate, we propose to perform the recognition by fusing gender information on the base of previous face recognition algorithm. The main idea of our proposed algorithm is that in the last recognition step of previous face recognition algorithm, we firstly obtain gender information of the input face image, and then perform recognition on the samples with the same gender as the input image in the face database. The experimental results demonstrate that our proposed algorithm has greatly improved the recognition rate.
Keywords: Face recognition gender recognition information fuse
论文图表:
引用
No.****
同行评议
共计0人参与
勘误表
融入性别信息的人脸识别方法
评论
全部评论