基于视听信息的自动年龄估计方法

Automatic Age Estimation Based on Visual and Audio Information

年龄是人的重要属性.近年来,自动估计用户年龄逐渐成为一个涉及模式识别、计算机视觉、语音识别、人机交互、机器学习等领域的活跃课题.其在现实世界中也有很多的实际应用,如法医学、电子商务、安全控制等等.日常生活中,人们往往可以很容易地根据视听信息(这里主要指人脸和语音)来判断一个人的年龄,原因在于人脸和语音是人的年龄信息的重要载体.同样的,人机交互系统可以根据人脸图像以及语音来自动进行年龄估计.主要介绍了基于视听信息进行年龄估计的应用领域所遇到的挑战以及现有的解决方案.详细介绍了基于视听信息的年龄估计所用到的主要模型、算法及其性能与特点,并且分析了自动年龄估计未来可能的发展趋势.

Age is an important attribute of human beings. In recent years, automatic estimations of the user＇s age have been becoming an active topic in pattern recognition, computer vision, voice recognition, human-computer interaction （HCI）, etc. It can be widely used in many real applications such as forensics, e-business, security, and so on. In daily life, people can easily estimate the age of a person according to some visual and audio information （here mainly refers to face and voice） because humans＇ faces and voices are important agents of their age. This paper introduces in detail the models, algorithms used in automatic age estimation based on visual and audio information, as well as their performance and characteristics. The possible future directions for the research in automatic age estimation are also discussed.