基于人物图像视觉特征的人物性格隐私分析

doi:10.11959/j.issn.1000-436x.2016223

摘要/Abstract

摘要：

社交媒体中的个性化推荐试图通过对公开在公众平台中的用户位置、图片等信息的分析得到用户的习惯、性格等，以提供个性化服务，但也因此给用户的隐私安全造成威胁。从隐私挖掘的角度，提出了一种基于人物图像的人物性格隐私分析方法。基于认知科学中的人物性格模型，提出了5类视觉特征，包括颜色、纹理、形状、伊顿对比和表情特征。实验表明，所提方法可以有效挖掘用户的性格隐私，特征选择实验进一步验证了所提出特征的有效性。

关键词: 人物图像, 视觉特征, 人物性格, 隐私

Abstract:

Personalized recommendation in social media attempts to analyze the habits and personality based on the loca-tion information and pictures from the users' comments on the public platform. This is very useful for personalized in-formation recommendation. But from perspective of privacy protection, this will affect the privacy security. A novel hu-man personality privacy analysis method based on portrait was proposed. By analyzing the personality model of the psy-chology, five visual features were proposed, including color features, texture features, shape features, Itten contrast fea-tures and expression features. Comprehensive experiments demonstrate the effectiveness of the proposed method. Further, the features analysis experiment show that the proposed features are very relevant to human personality privacy analyz-ing.

Key words: portrait, visual feature, personality, privacy

聂婕,黄磊,李臻,张健,闫艳,魏志强. 基于人物图像视觉特征的人物性格隐私分析[J]. 通信学报, 2016, 37(11): 129-136.

Jie NIE,Lei HUANG,Zhen LI,Jian ZHANG,Yan YAN,Zhi-qiang WEI. Human personality privacy analysis based on visual features[J]. Journal on Communications, 2016, 37(11): 129-136.

图/表 9

图1

表1

图2

表2

表3

表4

表5

图3

图4

参考文献 22

[1]	冯登国，张敏，李昊 . 大数据安全与隐私保护[J]. 计算机学报, 2014,37(1):246-258. FENG D G , ZHANG M , LI H . Big data security and privacy protec-tion[J]. Chinese Journal of Computers, 2014,37(1):246-258.
[2]	付岩，王耀威，王伟强，等． SVM 用于基于内容的自然图像分类和检索[J]. 计算机学报, 2003,26(10):1261-1265. FU Y , WANG Y W , WANG W Q , et al. Content-based natural image classification and retrieval using SVM[J]. Chinese Journal of Com-puters, 2003,26(10):1261-1265.
[3]	MACHAJDIK J , HANBURY A . Affective image classification using features inspired by psychology and art theory[C]// International Con-ference on Multimedia. 2010:83-92.
[4]	JIA J , WU S , WANG X , et al. Can we understand van gogh's mood?:learning to infer affects from images in social networks[C]// ACM In-ternational Conference on Multimedia. ACM, 2012:857-860.
[5]	王伟凝，蚁静缄，贺前华 . 可计算图像美学研究进展[J]. 中国图象图形学报, 2012,7(8):893-901. WANG W N , YI J J , HE Q H . Review for computational image aes-thetics[J]. Journal of Image and Graphics, 2012,7(8):893-901.
[6]	CHEN Y Y , HSU W H , LIAO H Y M , et al. Automatic training image ac-quisition and effective feature selection from community-contributed photos for facial attribute detection[J]. IEEE Trans Multimedia, 2013,15(6):1388-1399.
[7]	HATIA S , LUO J . A picture tells a thousand words—about you! User interest profiling from user generated visual content[C]// Signal Proc-ess, Special Issues on Big Data Meets Multimedia Analytics, 2015.
[8]	WANG X , JIA J , TANG J , et al. Modeling emotion influence in image social networks[J]. IEEE Trans Affect Comput, 2015,6(3):286-297.
[9]	NIE J , CUI P , YAN Y , et al. How your portrait impresses people?:inferring personality impressions from portrait contents[C]// Proceed-ings of the ACM International Conference on Multimedia. ACM, 2014:905-908.
[10]	YAN Y , NIE J , HUANG L , et al. Is your first impression reliable? trustworthy analysis using facial traits in portraits[C]// Proceedings of International Conference on MultiMedia Modeling. 2015:148-158.
[11]	VIOLA P , JONES M . Rapid object detection using a boosted cascade of simple features[J]. Proc Cvpr, 2001:1-511.
[12]	DE RAAD B . The big five personality factors: the psycholexical approach to personality[M]. Hogrefe ＆ Huber Publishers, 2000.
[13]	CRISTANI M , VINCIARELI A , SEGALIN C , et al. Unveiling the multimedia unconscious: implicit cognitive processes and multimedia content analysis[C]// ACM Multimedia. 2013:213-222.
[14]	DATTA R , JOSHI D , LI J , et al. Studying aesthetics in photographic images using a computational approach[C]// European Conference on Computer Vision. 2006:288-301.
[15]	JUPP P E . Directional statistics[M]. Wiley, New York, 2004.
[16]	VALDEZ P , MEHRABIAN A . Effects of color on emotions[J]. Journal of Experimental Psychology General, 1994,123(4):394-409.
[17]	ROBERT H , LINDA S . Computer and robot vision[M]. Addi-son-Wesley Longman Publishing Co., Inc., Boston, 1992.
[18]	TAMURA H , MORI S , YAMAWAKI T . Textural features correspond-ing to visual perception[J]. IEEE Transactions on Systems Man ＆Cybernetics, 1978,8(6):460-473.
[19]	DALAL N , TRIGGS B . Histograms of oriented gradients for human detection[C]// IEEE Conference on Computer Vision ＆ Pattern Recog-nition. 2013:886-893.
[20]	ITTEN J . The art of color : the subjective experience and objective rationale of color[M]. John Wiley, New York, 1973.
[21]	WANG W N , YU Y L , JIANG S M . Image retrieval by emotional semantics: a study of emotional space and feature extraction[C]// IEEE International Conference on Systems, Man and Cybernetics, 2006:3534-3539.
[22]	CHANG C C , LIN C J . LIBSVM: a library for support vector ma-chines[J]. ACM Transactions on Intelligent Systems ＆ Technology, 2011,2(3):389-396.

序号	名称	长度	描述
1	色调、饱和度、亮度	42	所有图像区域的（7个）在色调、饱和度和亮度3个通道的均值和方差
2	PAD特征	6	全图和服装区域在pleasure、arousal和dominance 3个维度的特征
3	灰度共生矩阵（GLCM）	20	全图、面部区域、发型区域和鼻嘴区域、服装区域的特征
4	Tamura特征	30	全图、面部区域、发型区域和鼻嘴区域、服装区域的特征
5	边缘直方图	84	所有图像区域（7个）的边缘进行梯度统计，将360°平均划分为12个单元
6	方向梯度直方图（HOG）	84	眼部区域（24）、眉毛区域（24）、鼻嘴区域（36）进行统计
7	伊顿对比特征	21	详见表2
8	表情特征	1	生气(anger)、厌恶(disgust)、害怕(fear)、微笑(smile)、大笑(laugh)、伤心(sadness)、惊讶(surprise)和中立(neutral)

对比特征	符号	区域对
色度对比	H_Itten	全图和服装，全图和面部，面部和鼻嘴部，面部和服装，面部和发型，发型和服装
亮度对比	L_Itten	全图和面部，面部和服装，眼部和鼻嘴部
饱和度对比	S_Itten	全图和面部，面部和服装，眼部和鼻嘴部
带权重的亮度对比	L_Itten_weighted	全图和面部，面部和服装，眼部和鼻嘴部
冷暖对比	U_Itten	全图和服装，全图和面部，面部和鼻嘴部，面部和服装，面部和发型，发型和服装

人格类型	有效样本总数	样本数量（+/?）
经验开放性(O)	1 876	564/1 312
尽责性(C)	1 543	873/670
外倾性(E)	1 679	742/973
宜人性(A)	1 849	321/1 528
情绪稳定性(N)	1 753	599/1 154

序号	特征名称	O	C	E	A	N
1	色调	●	●	●	●	●
2	饱和度		●	●		●
3	亮度		●	●		●
4	全图．PAD	●	●	●	●	●
5	服装．PAD	●	●	●	●
6	全图．GLCM		●	●
7	面部．GLCM		●	●
8	发型．GLCM		●	●
9	鼻嘴．GLCM		●	●		●
10	服装．GLCM	●	●	●	●
11	全图．Tamura		●	●	●
12	面部．Tamura		●	●	●	●
13	发型．Tamura		●	●
14	鼻嘴．Tamura		●	●
15	边缘直方图		●	●
16	眼部．HOG		●	●
17	眉毛．HOG		●	●		●
18	鼻嘴．HOG		●	●	●	●
19	全图和服装．伊顿色度	●	●	●	●	●
20	全图和面部．伊顿色度	●	●	●	●	●
21	面部和鼻嘴部．伊顿色度		●	●		●
22	面部和服装．伊顿色度	●	●	●	●	●
24	面部和发型．伊顿色度		●	●	●
25	发型和服装．伊顿色度		●	●		●
26	全图和面部．伊顿亮度	●	●	●	●
27	面部和服装．伊顿亮度		●	●	●
28	眼部和鼻嘴部．伊顿亮度	●	●	●	●
29	全图和面部．伊顿饱和度		●	●
30	面部和服装．伊顿饱和度		●	●
31	眼部和鼻嘴部．伊顿饱和度	●	●	●	●
32	全图和面部．伊顿亮度（权重）	●	●	●	●	●
33	面部和服装．伊顿亮度（权重）	●	●	●	●	●
34	眼部和鼻嘴部．伊顿亮度（权重）		●	●		●
35	全图和服装．伊顿冷暖	●	●	●	●	●
36	全图和面部．伊顿冷暖	●	●	●	●	●
37	面部和鼻嘴部．伊顿冷暖		●	●		●
38	面部和服装．伊顿冷暖		●	●		●
39	面部和发型．伊顿冷暖		●	●	●	●
40	发型和服装．伊顿冷暖	●	●	●	●	●
41	表情特征	●	●	●	●	●
	特征选择方法	G	P	P	G	G

人格类型	主观方法	本文方法	基准方法
经验开放性(O)	78.83%	61.75%	49.81%
尽责性(C)	81.75%	67.47%	62.15%
外倾性(E)	89.33%	74.75%	68.63%
宜人性(A)	85.15%	55.96%	52.33%
情绪稳定性(N)	75.18%	66.77%	64.16%