针对非限制性条件下的人脸识别存在自由度高、干扰因素复杂等技术难点,引入深度学习理论,提出了一种基于深度信念网络(DBNs)的非限制性人脸识别算法模型。基于相对熵稀疏性限制和dropout机制等方法,设计了优化算法。针对实际使用场合中样本量不足的问题,提出了一种混合DBNs模型,该模型采用CNNs深度卷积网络生成训练DBNs所需的模拟样本。标准人脸库下的实验结果表明,DBNs模型的平均识别率为97.0%,混合DBNs模型的平均识别率为90.3%,满足实际使用需求。
Abstract
To overcome the technical difficulties like high degrees of freedom and the complicated interference factors in unconstrained face recognition, an algorithm for unconstrained face recognition based on DBNs network is proposed with the adoption of deep learning theory. Based on relative entropy sparse restrictions and dropout mechanism, the optimization algorithm is designed. As for the problem of small sample in practice, an algorithm based on hybrid DBNs network model is proposed, which generates simulated samples with CNNs model to train the DBNs network. When tested by the standard face library, the experimental results show that the average recognition accuracy of DBNs and hybrid DBNs reach 97.0% and 90.3% respectively, which satisfy the practical using demand
关键词
计量学 /
人脸识别 /
深度信念网络 /
深度学习 /
小样本
Key words
metrology /
face recognition /
DBNs network /
deep learning /
small sample
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
［1］林妙真. 基于深度学习的人脸识别研究［D］.大连:大连理工大学,2013.
［2］刘锐. 基于人脸图像稠密匹配的身份识别技术研究［D］.合肥:中国科学技术大学,2014.
［3］胡伟,张少华,郭晓丽. 基于差异性稀疏表示的人脸识别算法［J］.计算机科学,2014,41(6):178-180.
［4］Shih P, Liu C. Face detection using discriminating feature analysis and support vector machine in video［J］. Pattern Recognition,2004,39(2):260-276.
［5］王莹,樊鑫,李豪杰,等. 基于深度网络的多形态人脸识别［J］.计算机科学,2015,42(9):61-65.
［6］Viola P, Jones M J. Robust Real-time Face Detection［J］. International Journal of Computer Vision,2004,57(2):137-154.
［7］章焱. 基于深度神经网络的桥牌识别系统研究［D］.厦门:厦门大学,2014.
［8］吴凤和. 基于计算机视觉测量技术的图像轮廓提取方法研究［J］.计量学报,2007,28(1):18-22.
［9］Prince S J D, Warrell J, Elder J H, et al. Tied Factor Analysis for Face Recognition across Large Pose Differences［J］. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2008, 30(6):970-984.
［10］Li B, Chang H, Shan S, et al. Low-resolution Face Recognition via Coupled Locality Preserving Mappings［J］. IEEE Signal Processing Letters, 2010, 17(1):20-23.
［11］熊娟,文桦. 基于图论萤火虫搜索算法的图像纹理特征提取研究［J］.计量学报,2016, 37(3): 255-259.
［12］Hinton G E. A Practical Guide to Training Restricted Boltzmann Machines［J］. Momentum, 2010, 9(1):599-619.
［13］由清圳. 基于深度学习的视频人脸识别方法［D］.哈尔滨:哈尔滨工业大学,2013.
［14］奚雪峰,周国栋. 基于Deep Learning的代词指代消解［J］.北京大学学报(自然科学版),2014,50(1):100-110.
［15］王宪保,李洁,姚明海,等.基于深度学习的太阳能电池片表面缺陷检测方法［J］.模式识别与人工智能,2014,27(06):517-523.
基金
国家自然科学基金(61471191);航空基金（20152052026）
PDF(800 KB)
