您的位置:山东大学 -> 科技期刊社 -> 《山东大学学报(工学版)》

山东大学学报(工学版) ›› 2016, Vol. 46 ›› Issue (2): 6-13.doi: 10.6040/j.issn.1672-3961.1.2015.160

• 机器学习与数据挖掘 • 上一篇    下一篇

融合二级评价指标的人脸图像质量评价方法

邹国锋1,傅桂霞1,李震梅1,李海涛1,王科俊2   

  1. 1. 山东理工大学电气与电子工程学院, 山东 淄博 255049;2. 哈尔滨工程大学自动化学院, 黑龙江 哈尔滨 150001
  • 收稿日期:2015-05-12 出版日期:2016-04-20 发布日期:2015-05-12
  • 作者简介:邹国锋(1984— ),男,山东泰安人,讲师,博士,主要研究方向为人脸检测与识别,生物特征识别与智能监控,模式识别理论及应用. E-mail:zgf841122@163.com
  • 基金资助:
    山东省自然科学基金博士基金资助项目(ZR2015FL029);国家自然科学基金青年科学基金资助项目(51407112)

Face image quality evaluation method based on the fusion of two level evaluation indexes

ZOU Guofeng1, FU Guixia1, LI Zhenmei1, LI Haitao1, WANG Kejun2   

  1. 1. College of Electrical and Electronic Engineering, Shandong University of Technology, Zibo 255049, Shandong, China;
    2. College of Automation, Harbin Engineering University, Harbin 150001, Heilongjiang, China
  • Received:2015-05-12 Online:2016-04-20 Published:2015-05-12

摘要: 针对姿态和光照对人脸的干扰,提出基于子区域直方图距离的人脸对称度评价,用于评估非对称光照和姿态对人脸质量的影响。提出针对含有人脸的原始图像质量的第一级评价与针对人脸有效区域的第二级评价相融合的评价策略,第一级评价的反馈信息能有效指导图像采集环境构建和改善,为后期人脸检测和识别提供优质图像源。主要评价指标包括:对比度、适宜度、对称度、清晰度、人脸有效区域面积等物理参数。实验结果表明了本研究提出的对称度评价方法和融合二级指标评价策略的可靠性与准确性。

关键词: 适宜度, 子区域直方图, 有效区域面积, 对比度, 人脸图像质量评价, 对称度, 清晰度

Abstract: To overcome the influence of face pose and illumination, the face symmetry degree evaluation method based on sub-regions histogram distance was proposed, which could be used to evaluate the effect on face quality of asymmetric illumination and pose. A novel evaluation strategy was proposed, which was the fusion of the first level evaluation index for the original natural image including face and the second level evaluation index for face effective area. The feedback information of the first level evaluation could effectively guide the construction and improvement of image acquisition environment, to provide high quality image source for the latter part of face detection and recognition. The main evaluation indexes included contrast degree, suitability degree, symmetry degree, clarity and effective area. Experiment results showed that the proposed symmetry degree evaluation method and the fusion strategy of two level evaluation indexes had good reliability and accuracy.

Key words: symmetry degree, suitability degree, sub-regions histogram distance, effective area, face image quality evaluation, contrast degree, clarity

中图分类号: 

  • TP391.41
[1] 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.
[2] ZOU G F, JIANG S M, ZHANG Y Y, et al. A novel coupled metric learning method and its application in degraded face recognition[C] //The 8th Chinese Conference on Biometric Recognition. Heidelberg, Germany:Springer, 2013:152-161.
[3] ZOU G F, ZHANG Y Y, JIANG S M, et al. An improved metric learning approach for degraded face recognition[J]. Mathematical Problems in Engineering, 2014:1-10.
[4] GAO X, LU W, TAO D, et al. Image quality assessment based on multi-scale geometric analysis[J]. IEEE Transaction on Image Processing, 2009, 18(7):1409-1423.
[5] 梁敏瑜,孙权森. 基于结构特征的图像质量评价模型[J].山东大学学报(工学版),2012,42(3):52-56. LIANG Minyu, SUN Quansen. An image quality assessment model based on structure feature[J]. Journal of Shandong University(Engineering Science), 2012, 42(3):52-56.
[6] 任玉玲,路文,徐红强,等. 一种基于Shearlet变换的图像质量客观评价方法[J]. 山东大学学报(工学版),2015,45(3):15-21. REN Yuling, LU Wen, XU Hongqiang, et al. An image quality assessment method based on Shearlet transform[J]. Journal of Shandong University(Engineering Science), 2015, 45(3):15-21.
[7] DAYRON R R, HEYDI M V, EDEL G R. An illumination quality measure for face recognition[C] //2010 International Conference on Pattern Recognition(ICPR2010). Istanbul, Turkey: IEEE, 2010:1477-1480.
[8] AYMAN A, MARY A H, THIRIMACHOS B. Quality metrics for practical face recognition[C] //2012 International Conference on Pattern Recognition(ICPR 2012). Tsukuba, Japan: IEEE, 2012:3103-3107.
[9] JONATHON PHILLIPS P, ROSS BEVERIDGE J, David Bolme, et al. On the existence of face quality measures[C] //IEEE 6th International Conference on Biometrics: Theory, Applications and Systems. Washington, United States: IEEE, 2013:1-8.
[10] 胡杨庆,童卫清.人脸图像质量评价法[J].现代计算机,2007(12):43-45. HU yangqing, TONG weiqing. Assessment of human-face image quality[J]. Modern Computer, 2007(12):43-45.
[11] 闫日亮, 张会林, 黄金钰. 基于信息熵和Harris 算法的人脸图像质量评价[J].计算机安全,2011(9):11-12. YAN Riliang, ZHANG Huilin, HUANG Jinyu. Based on the information entropy and harris algorithm of image quality assessment[J]. Computer Security, 2011(9):11-12.
[12] 高修峰,张培仁,李子青.人脸图像质量评估标准[J].小型微型计算机系统,2009,30(1):95-99. GAO Xiufeng, ZHANG Peiren, LI Ziqing. Standardization of face image sample quality[J]. Journal of Chinese Computer Systems, 2009, 30(1):95-99.
[13] 高修峰. 人脸图像质量评估标准方法研究[D].合肥:中国科学技术大学,2008. GAO Xiufeng. The research on face image quality assessment[D]. Hefei:University of Science and Technology of China, 2008.
[14] 杨飞, 苏剑波.一种基于倒谱的人脸图像清晰度评价方法[J]. 光电子.激光,2009,20(10):1357-1360. YANG Fei, SU Jianbo. A cestrum-based clarity assessment method for face images[J]. Journal of Optoelectronics Laser, 2009, 20(10):1357-1360.
[15] LU Wen, GAO Xinbo, TAO Dacheng, et al. A wavelet-based image quality assessment method[J]. Journal of Wavelets, Multiresolution, and Information Processing, 2008, 6(4):541-551.
[16] 肖宾杰. 基于图像质量加权的D-S证据理论多生物特征融合识别[J]. 计算机应用,2012,32(1):264-268. XIAO Binjie. Multi-biometric feature fusion identification based on D-S evident theory of image quality with different weights[J]. Journal of Computer Applications, 2012, 32(1):264-268.
[17] 吴限,刘崎峰. 人脸检测建模源照片筛选方法[J].应用科技,2015,42(1):33-35. WU Xian, LIU Qifeng. A method of selecting source data of picture in the modeling of face detection[J]. Applied Science and Technology, 2015, 42(1):33-35.
[18] 高新波,路文.视觉信息质量评价方法[M]. 西安: 西安电子科技大学出版社,2010.
[19] 李奇,冯华君,徐之海,等. 数字图象清晰度评价函数研究[J].光子学报, 2002, 31(6):736-738. LI Qi, FENG Huajun, XU Zhihai, et al. Research on measurement function of digital image definition[J]. Acta Photonica Sinica, 2002, 31(6):736-738.
[20] 王鸿南,钟文,汪静,等. 图像清晰度评价方法研究[J]. 中国图象图形学报,2004,9(7):828-831. WANG Hongnan, ZHONG Wen, WANG Jing, et al. Research of measurement for digital image definition[J]. Journal of Image and Graphics, 2004, 9(7):828-831.
[21] WANG Z, BOVIK A C, SHEIKH H R, et al. Image quality assessment: from error visibility to structural similarity[J]. IEEE Transactions on Image Processing, 2004, 13(4):600-612.
[22] 唐文剑. 正面人脸图像质量评价方法研究[D].西安:西安电子科技大学,2012. TANG Wenjian. Frontal face image quality assessment [D]. Xi'an: Xidian University, 2012.
[23] 章毓晋. 图像处理基础教程[M]. 北京:电子工业出版社,2012.
[24] 吕萌,苏红旗,刘启春,等. 一种新的自适应边缘提取微分算子[J]. 数据采集与处理,2011,26(1):106-111. L(¨overU)Meng, SU Hongqi, LIU Qichun, et al. New adaptive differential operator for edge detection[J]. Journal of Data Acquisition and Processing, 2011, 26(1):106-111.
[25] 中国科学院计算机技术研究所. CAS-PEAL-R1人脸库[EB/OL].(2015-01-20)[2015-03-20]. http://www.ict.ac.cn/jszy/jsxkzlxk/jsxk/200707/t20070706-2179-538.html.
[1] 曹红根1,袁宝华1,朱辉生2. 结合对比度信息与LBP的分块人脸识别[J]. 山东大学学报(工学版), 2012, 42(4): 29-34.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 沈晓晶, 陈明, 池涛. 多Agent水质监控系统中的信息融合算法[J]. 山东大学学报(工学版), 2014, 44(4): 39 -45 .
[2] 王立宏,李强. 旅行商问题的一种选择性集成求解方法[J]. 山东大学学报(工学版), 2016, 46(1): 42 -48 .
[3] 范德斌,邓长寿,袁斯昊,谭旭杰,董小刚. 基于MapReduce模型的分布式粒子群算法[J]. 山东大学学报(工学版), 2016, 46(6): 23 -30 .
[4] 王祎珺,张晖,李波,杨春明,赵旭剑. 一种基于话题演化的意见领袖发现方法[J]. 山东大学学报(工学版), 2016, 46(2): 35 -42 .
[5] 刘大琨,谭晓阳. 基于流形的约束局部模型拟合[J]. 山东大学学报(工学版), 2016, 46(3): 31 -36 .
[6] 刘志军. 基于复合混沌与仿射变换的彩色图像加密算法[J]. 山东大学学报(工学版), 2016, 46(4): 1 -8 .
[7] 吉兴全,韩国正,李可军,傅荣荣,朱仰贺. 基于密度的改进K均值聚类算法在配网区块划分中的应用[J]. 山东大学学报(工学版), 2016, 46(4): 41 -46 .
[8] 刘斌,宋锐,柴汇. 基于虚拟模型和加速度规划的腿部缓冲策略[J]. 山东大学学报(工学版), 2016, 46(6): 69 -75 .
[9] 易云飞,苗剑,林郭隆,殷智. 基于牛顿力学和博弈论模型的粒子网络优化算法[J]. 山东大学学报(工学版), 2017, 47(1): 28 -36 .
[10] 万利,王春河,王琦,李术才,邵行,江贝,孙会彬,秦乾. 超大断面隧道软弱围岩控制机制及应用[J]. 山东大学学报(工学版), 2017, 47(1): 59 -67 .