Nguyen Dat Tien, Park Kang Ryoung
Division of Electronics and Electrical Engineering, Dongguk University, 30, Pildong-ro 1-gil, Jung-gu, Seoul 100-715, Korea.
Sensors (Basel). 2016 Jan 27;16(2):156. doi: 10.3390/s16020156.
Gender information has many useful applications in computer vision systems, such as surveillance systems, counting the number of males and females in a shopping mall, accessing control systems in restricted areas, or any human-computer interaction system. In most previous studies, researchers attempted to recognize gender by using visible light images of the human face or body. However, shadow, illumination, and time of day greatly affect the performance of these methods. To overcome this problem, we propose a new gender recognition method based on the combination of visible light and thermal camera images of the human body. Experimental results, through various kinds of feature extraction and fusion methods, show that our approach is efficient for gender recognition through a comparison of recognition rates with conventional systems.
性别信息在计算机视觉系统中有许多有用的应用，如监控系统、统计商场中的男性和女性数量、限制区域的访问控制系统或任何人机交互系统。在大多数先前的研究中，研究人员试图通过使用人脸或身体的可见光图像来识别性别。然而，阴影、光照和一天中的时间会极大地影响这些方法的性能。为了克服这个问题，我们提出了一种基于人体可见光和热成像相机图像相结合的新的性别识别方法。通过各种特征提取和融合方法的实验结果表明，通过与传统系统的识别率进行比较，我们的方法在性别识别方面是有效的。
