
肖志涛
1
,伊 靓
1
,李月龙
2
,张 芳
1
,耿 磊
1
,吴 骏
1
1. 3003872. 300387
针对传统方法合成的正面人脸图像中信息丢失和变形的问题, 提出了一种基于筛选评估准则的非正面人
脸图像合成方法. 人脸筛选评估准则融合了脸部对称性、 正脸差异水平和人脸水平扭转角度 3 方面 信息,
其中人脸水平扭转角度利用细节上的眼部信息来评价人脸的正面水平, 而脸部对称性和正脸差异水平分
别对人脸的左右和垂直方向进行整体评价, 综合这三方面信息可有效地排除低质量侧脸图像对合成正脸
图像的干扰. 首先进行标记点检测跟踪,然后基于此对同一人的多幅侧脸图像进行筛选,最后进行插值运
算合成正面人脸, 并在 FERET 图像库中对该方法进行实验验证. 结果表明:通过本文筛选准则可有效滤除
合成中低质量、强干扰的侧脸图像,可降低姿态问题对人脸识别精度的干扰,最终合成精确逼近真实正面
人脸的合成图像.
非正面人脸合成;筛选评估准则;人脸识别
TP391 A 1671-024X201502-0069-06
XIAO Zhi-tao
1
YI Jing
1
LI Yue-long
2
ZHANG Fang
1
GENG Lei
1
WU Jun
1
1. School of Electronics and Information Engineering Tianjin Polytechnic University Tianjin 300387 China 2. Schoo l of
Computer Science and Software Engineering Tianjin Polytechnic University Tianjin 300387 China
To solve the problem of information missing and deformation of the traditional synthesized frontal face image
,
a
non-frontal face image synthesis method based on source sc reening criteria is proposed. The source screening
criteria combines the facial symmetry
,
the frontal face image differences level and the horizontal rotating angle.
The horizontal rotating angle uses the details of the eye information to assess the level of the front face. The left-
to-right and vertical direction is separately asses s ed according to the facial symmetry and the frontal face image
differences level. Combining these three aspects can availably exclude interference of low -quality non -frontal
face image in synthesizing the frontal face. Firstly
,
marking points are detected and tracked. Then
,
multi-view
face images of same person are screened to obtain the high-quality
,
low-interference non-frontal face images as
the best synthetic input source images. Finally
,
frontal face images are synthesized by ca lculated interpolation.
Experimental findings on the FERET databases demonstrate that the so urc e screening criteria can availably
screen out low-quality
,
strong-interference non-frontal fac e images and reduce the impact of pose questions on
the precision of face recognition algorithm effectively. The synthesized frontal view face can approximate the
ground truth frontal view face.
non-frontal face image synthesis source screening criteria face recognition
2014-12-15 613021271132619820120805
1971 . E-mailxiaozhitao@tjpu.edu.cn.
天 津 工 业 大 学 学 报
第
34
卷 第
2
期
2015
年
4
月
Vol.34 No.2
April 2015
.
[1]
.
[2]
.
. Bl anz
[3]