基于吸引模式的局部二阶梯度轮廓人脸识别算法

doi:10.11959/j.issn.1000-0801.2021140

电信科学 ›› 2021, Vol. 37 ›› Issue (7): 96-106.doi: 10.11959/j.issn.1000-0801.2021140

基于吸引模式的局部二阶梯度轮廓人脸识别算法

叶学义, 钱丁炜, 应娜, 王涛

杭州电子科技大学模式识别与信息安全实验室，浙江杭州 310018

修回日期:2021-07-10 出版日期:2021-07-20 发布日期:2021-07-01
作者简介:叶学义（1973- ），男，博士，杭州电子科技大学模式识别与信息安全实验室副教授，主要研究方向为图像处理、模式识别与信息隐藏等
钱丁炜（1996- ），男，杭州电子科技大学模式识别与信息安全实验室硕士生，主要研究方向为图像处理
应娜（1978- ），女，博士，杭州电子科技大学讲师，主要研究方向为语音信号处理与图像处理等
王涛（1990- ），男，杭州电子科技大学模式识别与信息安全实验室硕士生，主要研究方向为图像处理
基金资助:
国家自然科学基金资助项目(60802047);国家自然科学基金资助项目(60702018);国家自然科学重点基金资助项目(U19B2016)

Face recognition algorithm based on attractive local second gradient contours

Xueyi YE, Dingwei QIAN, Na YING, Tao WANG

Laboratory of Pattern Recognition ＆ Information Security，Hangzhou Dianzi University，Hangzhou 310018，China

Revised:2021-07-10 Online:2021-07-20 Published:2021-07-01
Supported by:
The National Natural Science Foundation of China(60802047);The National Natural Science Foundation of China(60702018);Major Program of the National Natural Science Foundation of China(U19B2016)

摘要/Abstract

摘要：

针对现有LDP类算法在特征提取有效性和特征编码稳定性之间难以平衡的问题，提出一种吸引局部二阶梯度轮廓（ALSGC）模式，以提升人脸识别性能。首先，利用Kirsch算子计算人脸的邻域边缘响应图；其次，引入吸引描述子，参考边缘响应图的局部、全局平均灰度值和邻域中心灰度值完成局部吸引模式编码；再次，遍历整幅图像，得到人脸ALSGC特征图并对ALSGC特征图分块分别计算，得到各个分块中不同模式的统计直方图；最后，级联所有分块的统计直方图后生成对应的特征向量，以支持向量机完成分类识别。所提算法克服了LBP、LDP、LDN等算法提取一阶特征有效性的不足，以及DLDP、CSLDP、GCSLDP等算法提取的二阶特征对表情、姿态、饰物遮挡、光照、随机噪声等变化敏感的缺点，较好地实现了特征提取有效性与特征编码稳定性的平衡，兼顾了识别率和稳健性。

关键词: 局部吸引模式, 二阶梯度轮廓, Kirsch算子, 特征提取, 人脸识别

Abstract:

In order to solve the problem that traditional LDP algorithms is difficult to balance the effectiveness of feature extraction and the stability of feature encoding, an attractive local second gradient contours (ALSGC) face feature extraction algorithm was proposed.Firstly, the Kirsch operator was used to calculate the neighborhood edge response image of the face image.Secondly, an attraction descriptor was introduced and the local average gray value and global average gray value of the edge response image and the neighborhood center gray value was combined to complete the local attraction pattern encoding.The entire image was traversed to get the ALSGC face feature map, the ALSGC feature map was divided to blocks and the statistical histograms of different patterns for each block was obtained by calculation.Finally, the statistical histograms of all the blocks were cascaded to generate the corresponding feature vector and support vector machine was used to complete classification and recognition.The proposed algorithm not only overcame the insufficient effectiveness of LBP, LDP, LDN and other algorithms who extract first-gradient features, but also reduced the sensitivity to changes in expression, posture, occlusion, lighting and random noise of methods such as DLDP, CSLDP, GCSLDP who extracting second-gradient features.It better achieved the balance between the effectiveness of feature extraction and the stability of feature encoding, and took into account the recognition rate and robustness.

Key words: local attractive pattern, second gradient contours, Kirsch operator, feature extraction, face recognition

中图分类号:

TP391

叶学义, 钱丁炜, 应娜, 王涛. 基于吸引模式的局部二阶梯度轮廓人脸识别算法[J]. 电信科学, 2021, 37(7): 96-106.

Xueyi YE, Dingwei QIAN, Na YING, Tao WANG. Face recognition algorithm based on attractive local second gradient contours[J]. Telecommunications Science, 2021, 37(7): 96-106.

图/表 15

图1

图2

图3

图4

图5

图6

图7

表1

表2

表3

图8

表4

图9

图10

表5

参考文献 17

[1]	JAYARAMAN U , GUPTA P , GUPTA S ,et al. Recent development in face recognition[J]. Neurocomputing, 2020(408): 231-245.
[2]	ROUSSEEUW P J , HUBERT M . Anomaly detection by robust statistics[J]. Wiley Interdisciplinary Reviews:Data Mining and Knowledge Discovery, 2018,8(2): e1236.
[3]	KIM C , KLABJAN D . A simple and fast algorithm for L1-norm kernel PCA[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020,42(8): 1842-1855.
[4]	YE Q L , ZHAO H H , FU L Y ,et al. Underlying connections between algorithms for nongreedy LDA-L1[J]. IEEE Transactions on Image Processing, 2018,27(5): 2557-2559.
[5]	OJALA T , VALKEALAHTI K , OJA E ,et al. Texture discrimination with multidimensional distributions of signed gray-level differences[J]. Pattern Recognition, 2001,34(3): 727-739.
[6]	JABID T , KABIR M H , CHAE O . Local Directional Pattern (LDP) for face recognition[C]// Proceedings of 2010 Digest of Technical Papers International Conference on Consumer Electronics (ICCE). Piscataway:IEEE Press, 2010: 329-330.
[7]	RIVERA A , ROJAS CASTILLO J , CHAE O O . Local directional number pattern for face analysis:face and expression recognition[J]. IEEE Transactions on Image Processing, 2013,22(5): 1740-1752.
[8]	李照奎, 丁立新, 王岩 ,等. 基于差值局部方向模式的人脸特征表示[J]. 软件学报, 2015,26(11): 2912-2929.
	LI Z K , DING L X , WANG Y ,et al. Face feature representation based on difference local directional pattern[J]. Journal of Software, 2015,26(11): 2912-2929.
[9]	杨恢先, 翟云龙, 蔡勇勇 ,等. 基于中心对称梯度幅值相位模式的单样本人脸识别[J]. 光电子·激光, 2015,26(5): 969-977.
	YANG H X , ZHAI Y L , CAI Y Y ,et al. Face recognition based on center-symmetric gradient magnitude and phase pattern[J]. Journal of Optoelectronics·Laser, 2015,26(5): 969-977.
[10]	王晓华, 李瑞静, 胡敏 ,等. 融合局部特征的面部遮挡表情识别[J]. 中国图象图形学报, 2016,21(11): 1473-1482.
	WANG X H , LI R J , HU M ,et al. Occluded facial expression recognition based on the fusion of local features[J]. Journal of Image and Graphics, 2016,21(11): 1473-1482.
[11]	KHADIRI I , KAS M , EL MERABET Y ,et al. Repulsive-and-attractive local binary gradient contours:New and efficient feature descriptors for texture classification[J]. Information Sciences, 2018,467: 634-653.
[12]	杨恢先, 刘建, 张孟娟 ,等. 双空间局部方向模式的人脸识别[J]. 中国图象图形学报, 2017,22(11): 1493-1502.
	YANG H X , LIU J , ZHANG M J ,et al. Face recognition based on double-space local directional pattern[J]. Journal of Image and Graphics, 2017,22(11): 1493-1502.
[13]	SHABAT A M , TAPAMO J R . A comparative study of the use of local directional pattern for texture-based informal settlement classification[J]. Journal of Applied Research and Technology, 2017,15(3): 250-258.
[14]	CHEN L , LI Q , JIANG J J . Noisy practical facial super-resolution method via deformable constrained model with small dataset[J]. Multimedia Tools and Applications, 2020,79(3/4): 2577-2600.
[15]	HJOUJI A , EL-MEKKAOUI J ,, JOURHMANE M ,et al. New set of non-separable orthogonal invariant moments for image recognition[J]. Journal of Mathematical Imaging and Vision, 2020,62(4): 606-624.
[16]	ZHANG G Y , ZOU W B , ZHANG X J ,et al. Singular value decomposition based virtual representation for face recognition[J]. Multimedia Tools and Applications, 2018,77(6): 7171-7186.
[17]	王鹏, 叶学义, 王涛 ,等. 双偏差双空间局部方向模式的人脸识别[J]. 计算机工程与应用, 2021,57(4): 91-99.
	WANG P , YE X Y , WANG T ,et al. Face recognition based on double variation and double space local directional pattern[J]. Computer Engineering and Applications, 2021,57(4): 91-99.

算法	每类随机选择样本数k
算法	1	2	3	4
LBP	96.61%	98.76%	99.32%	99.51%
LDP	96.54%	98.59%	99.37%	99.56%
LDN	96.34%	98.67%	99.47%	99.67%
DLDP	96.43%	98.31%	98.89%	99.06%
CSLDP	95.21%	98.13%	98.91%	99.11%
GCSLDP	96.87%	98.49%	99.2%	99.37%
本文	96.88%	99.02%	99.73%	99.94%

算法	每类随机选择样本数k
算法	1	2	3	4	5
LBP	54.49%	70.98%	78.76%	84.38%	87.13%
LDP	55.92%	70.92%	79.29%	85.91%	88.07%
LDN	57.27%	73.9%	81.45%	86.26%	89.56%
DLDP	56.52%	74.13%	82.91%	89.84%	92.03%
CSLDP	58.72%	73.73%	81.37%	87.51%	90.93%
GCSLDP	58.82%	74.36%	82.9%	88.34%	91.24%
本文	65.45%	79.33%	87.61%	93.07%	96.81%

算法	每类随机选择样本数k
算法	3	4	5	6
LBP	51.49%	57.93%	66.82%	73.28%
LDP	53.32%	59.4%	67.78%	74.15%
LDN	59.78%	65.43%	73.6%	79.31%
DLDP	59.31%	65.34%	73.28%	78.45%
CSLDP	67.29%	74.91%	80.28%	84.63%
GCSLDP	71.73%	78.97%	84.82%	88.21%
本文	72.85%	80.86%	86.31%	90.52%

算法	每类随机选择样本数k
算法	3	4	5	6	7
LBP	82.07%	87.89%	91.25%	94.17%	95.08%
LDP	80.53%	87.08%	90.52%	93.27%	95.13%
LDN	82.48%	88.68%	91.77%	94.43%	96.12%
DLDP	81.42%	87.21%	90.63%	93.15%	94.46%
CSLDP	81.46%	86.23%	89.83%	92.22%	93.37%
GCSLDP	79.35%	84.35%	88.15%	91.34%	93.52%
本文	83.5%	90.5%	93.83%	95.65%	97.29%

算法	添加椒盐噪声比例
算法	0	10%	20%	30%	40%	δ
LBP	80.27%	76.7%	71.02%	61.87%	48.9%	39.08%
LDP	76.44%	76.15%	72.14%	69.15%	66.49%	13.02%
LDN	79.27%	77.06%	72.73%	65.91%	57.56%	27.39%
DLDP	79.62%	77.1%	73.14%	64.77%	58.28%	26.80%
CSLDP	71.57%	65%	62.97%	61.55%	60.57%	15.37%
GCSLDP	73.06%	66.87%	59.97%	50.17%	42.16%	42.29%
本文	80.89%	79.84%	78.17%	77.59%	73.26%	9.43%

基于吸引模式的局部二阶梯度轮廓人脸识别算法

Face recognition algorithm based on attractive local second gradient contours

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 15

参考文献 17

相关文章 12

Metrics

推荐阅读 0

[1]	尹良, 林睿, 王晓雷, 姚宇亮, 周林, 何元. 基于频谱形状的低复杂度雷达信号分类[J]. 电信科学, 2022, 38(1): 25-35.
[2]	李悦, 钱亚冠, 关晓惠, 李蔚, 王滨, 顾钊铨. 面向人脸识别的口罩区域修复算法[J]. 电信科学, 2021, 37(8): 66-76.
[3]	蔡鑫. 基于Bert模型的互联网不良信息检测[J]. 电信科学, 2020, 36(11): 121-126.
[4]	朱琳,赵娟,王伊婷,冯俊兰,邓超. 移动通信网络投诉热点问题智能预警方法[J]. 电信科学, 2019, 35(5): 17-24.
[5]	葛凯强, 陈铁明. 人机交互安全攻防综述[J]. 电信科学, 2019, 35(10): 100-116.
[6]	唐彪,金炜,符冉迪,龚飞. 多稀疏表示分类器决策融合的人脸识别[J]. 电信科学, 2018, 34(4): 31-40.
[7]	王潇,张荣,郭立君. 顽健合成图像篡改检测及定位[J]. 电信科学, 2017, 33(8): 120-129.
[8]	龚飞,金炜,朱珂晴,符冉迪,曹燕. 采用双字典协作稀疏表示的光照及表情顽健人脸识别[J]. 电信科学, 2017, 33(3): 52-58.
[9]	陆寄远,黄承慧,侯昉,李斌. 基于Hadoop云计算平台的图像分类与标注[J]. 电信科学, 2014, 30(2): 65-69.
[10]	魏莎莎,陆慧娟,金伟,李超. 基于云平台的互信息最大化特征提取方法研究[J]. 电信科学, 2013, 29(10): 38-42.
[11]	闫云斌,全厚德,崔佩璋. 跳频通信中阻塞干扰样式的自动识别[J]. 电信科学, 2013, 29(1): 103-108.
[12]	闫云斌,全厚德,崔佩璋. 跳频通信中阻塞干扰样式的自动识别[J]. 电信科学, 2012, 28(12): 46-51.