轻量级智能终端人脸识别系统研究与实现

doi:10.11959/j.issn.1000-436x.2015294

摘要/Abstract

摘要：

针对传统人脸识别算法不能有效适用于智能移动终端的问题，提出一种基于经典SIFT算法的特征加权分簇匹配的轻量级改进方案，该方案能自动学习、自适应添加可靠的测试样本到训练样本空间，具有合理划分和科学权值分配特性，使该方案在识别率和运行时间上都有提高。改进算法分别在 ORL 人脸库和 Yale 人脸库做了测试，相对于经典SIFT算法识别率提升了6.13%和14.11%，运行效率提升了9.1%和4.7%。同时按照Zhou的测试方法，在ORL人脸库识别率达到74.05%，比PCA、LBP等经典算法都有明显的提升,并在Android智能终端中对识别方案做了实现，实验数据验证了改进算法在 Android 系统的可用性，最后提出一种基于云架构的改进方案。

关键词: 人脸识别, 轻量级, 智能终端, SIFT, 检测区域划分

Abstract:

In order to solve this problem that traditional face recognition scheme was not efficiently suitable to intelligent terminal scene.An improved lightweight scheme of feature weighted clustering matching based on SIFT was presented.The scheme can learn automatically and adaptively add test samples to the training sample space.Reasonable division and scientific distribution of weight make this scheme has improved at run time and recognition rate.The improved algorithm has been tested in the ORL face database and Yale face database,compared with the classical SIFT algorithm the recognition rate improved by 6.13% and 14.11%,the running efficiency increased 9.1% and 4.7%.At the same time,in accordance with the test method by Zhou,in the ORL face database,the recognition rate was up to 74.05%,significantly improved than PCA,LBP and other classical algorithm.The algorithm in Android terminal is implemented,and the improved algorithm is verified to be available in the Android system by experiments.Finally,an improved scheme was proposed based on the cloud architecture.

Key words: face recognition, lightweight, intelligent terminal, SIFT, detection zoning

马国峻,周海东. 轻量级智能终端人脸识别系统研究与实现[J]. 通信学报, 2015, 36(Z1): 149-156.

Guo-jun MA,Hai-dong ZHOU. Research and implementation of intelligent terminal lightweight face recognition system[J]. Journal on Communications, 2015, 36(Z1): 149-156.

图/表 8

图1

图2

图3

表1

表2

表3

表4

图4

参考文献 14

[1]	LI B , WU S , ZENG F T . Implementation of face identification in intelligent mobile telephone[J]. Computer Engineering, 2006,32(7): 272-274.
[2]	LI W . Implementation of face identification in intelligent mobile telephone[J]. Computer Technology and Development, 2008,18(1): 161-163.
[3]	HU X P , CHEN D S , CHEN W B . Design and implementation of face detection system on smart phone[J]. Computer Engineering and Design, 2010,31(3): 672-675.
[4]	CHEN Y N , CHEN X G . A novel identify recognition system based on mobile phone and face recognition[J]. Computer Application Software, 2011,28(3): 77-79.
[5]	AHONEN T , HADID A,PIETIK?INEN M . Face Recognition With Local Binary Patterns[M]. Computer vision-eccv 2004. Springer Berlin Heidelberg, 2004. 469-481.
[6]	LOWE D G . Distinctive image features from scale-invariant keypoints[J]. International Journal of Computer Vision, 2004,60(2): 91-110.
[7]	LOWE D G . Object recognition from local scale-invariant features[A]. The Proceedings of the Seventh IEEE International Conference on Computer vision[C]. 1999. 1150-1157.
[8]	CAO S . A fast SURF way for human face recognition with cell similarity[A]. 2011 6th IEEE Conference on Industrial Electronics and Applications (ICIEA)[C]. 2011. 166-169.
[9]	ZHOU Z M , YU S Y , ZHANG Z R . An algorithm for face recognition based on sift descriptor[J]. Journal of Image and Graphics, 2008,13(10): 1882-1885.
[10]	LUO J , SHI Y X , DUAN D Y . Face recognition method based on SIFT feature[J]. Computer Engineering, 2010,36(13): 173-174.
[11]	HUANG A . Similarity measures for text document clustering[A]. Proceedings of the Sixth New Zealand Computer Science Research Student Conference (NZCSRSC2008)[C]. Christchurch,New Zealand, 2008. 49-56.
[12]	STREHL A , GHOSH J , MOONEY R . Impact of similarity measures on web-page clustering[A]. Workshop on Artificial Intelligence for Web Search (AAAI 2000)[C]. 2000. 58-64.
[13]	ALY M . Face Recognition Using SIFT Features[R]. CNS/Bi/EE Report, 2006. 186-193.
[14]	ZHAO R Z . Study on Face Recognition Based on Single-sample Training Set[D]. Chongqing:Chongqing University, 2012.

字段名	数据类型	描述
_id	整型	人脸数据表的主键，记录默认编号
name	字符型	姓名
major	字符型	专业
tutor	字符型	导师
picId1	字符型	图片名1，可以构造出图片的本地存储路径和描述子的本地存储路径
picId2	字符型	图片名2
sim	双精度浮点型	图片1与图片2的相似度

算法	识别率	测试时间/s
经典SIFT	0.602 2	306
仅划分7块（无权值）	0.66	293
L-SIFT(本文算法)	0.687 2	292
经典SIFT	0.741 6	602
仅划分7块（无权值）	0.748 8	547
L-SIFT(本文算法)	0.787 1	547

算法	识别率
PCA	0.663 9
2DPCA	0.697 2
LBP	0.547 2
SIFT	0.708 7
L-SIFT(本文算法)	0.740 5

项目	测试结果
注册速度/ms	746
注册失败率	0
误识率	0.24
拒识率	0.34
识别率	0.66
识别速度/ms	1 323

[1]	刘帅, 关杰, 胡斌, 马宿东. 基于MILP的轻量级密码算法ACE的差分分析[J]. 通信学报, 2023, 44(1): 39-48.
[2]	王振宇, 郭阳, 李少青, 侯申, 邓丁. 面向轻量级物联网设备的高效匿名身份认证协议设计[J]. 通信学报, 2022, 43(7): 49-61.
[3]	殷新春, 王梦宇, 宁建廷. 轻量级可搜索医疗数据共享方案[J]. 通信学报, 2022, 43(5): 110-122.
[4]	蒋梓龙, 金晨辉. Saturnin算法的不可能差分分析[J]. 通信学报, 2022, 43(3): 53-62.
[5]	李玮, 汪梦林, 谷大武, 李嘉耀, 蔡天培, 徐光伟. 轻量级密码算法TWINE的唯密文故障分析[J]. 通信学报, 2021, 42(3): 135-149.
[6]	张炳, 文峥, 赵宇轩, 王苧, 任家东. 双粒度轻量级漏洞代码切片方法评估模型[J]. 通信学报, 2021, 42(11): 233-241.
[7]	闫宏强,王琳杰. 物联网中认证技术研究[J]. 通信学报, 2020, 41(7): 213-222.
[8]	谢敏,李嘉琪,田峰. FeW的差分故障攻击[J]. 通信学报, 2020, 41(4): 143-149.
[9]	肖斌,景如霞,毕秀丽,马建峰. 基于分组SIFT的图像复制粘贴篡改快速检测算法[J]. 通信学报, 2020, 41(3): 62-70.
[10]	谢绒娜,李晖,史国振,郭云川. 基于属性轻量级可重构的访问控制策略[J]. 通信学报, 2020, 41(2): 112-122.
[11]	赵羽,杨洁,刘淼,孙金龙,桂冠. 面向视频监控基于联邦学习的智能边缘计算技术[J]. 通信学报, 2020, 41(10): 109-115.
[12]	武小年,李迎新,韦永壮,孙亚平. GRANULE和MANTRA算法的不可能差分区分器分析[J]. 通信学报, 2020, 41(1): 94-101.
[13]	谢敏,田峰,李嘉琪. TWINE算法的相关密钥不可能飞来去器攻击[J]. 通信学报, 2019, 40(9): 184-192.
[14]	李玮,吴益鑫,谷大武,李嘉耀,曹珊,汪梦林,蔡天培,丁祥武,刘志强. SIMON轻量级密码算法的唯密文故障分析[J]. 通信学报, 2019, 40(11): 122-137.
[15]	谭郁松,李荣振,吴庆波,张建锋,张尧学. 基于超算环境的面向多租户的轻量级虚拟HPC集群的设计与实现[J]. 通信学报, 2017, 38(Z2): 56-66.