ISSN 2394-3777 (Print)
ISSN 2394-3785 (Online)
Available online at www.ijartet.com
International Journal of Advanced Research Trends in Engineering and Technology (IJARTET)
Vol. 6, Issue 4, April 2019
All Rights Reserved © 2019 IJARTET 29
Image Processing Methods for 3D Face
Recognition
D. Malathi
1
, A.Mathangopi
2
, Dr. D. Rajini Girinath
3
P.G Student, Department of CSE, Sri Muthukumaran Institute of Technology, Chennai, India
1
Asst.Prof, Department of CSE, Sri Muthukumaran Institute of Technology, Chennai, India
2
Professor, Department of CSE, Sri Muthukumaran Institute of Technology, Chennai, India
3
Abstract: we present a novel automatic framework to perform 3D face recognition. The proposed method uses a Simulated
Annealing-based approach (SA) for range image adjustment with the Surface Interpenetration Measure (SIM), as similarity
measure, in order to match two face images. A modified SA approach is proposed taking advantage of constant face regions
to better handle facial expressions. The authentication score is obtained by combining the SIM values corresponding to the
matching of four different face regions: circular and elliptical areas around the nose, forehead and the entire face region.
Comprehensive experiments were performed on the FRGC v2 database, the largest available database of 3D face images.
By using all the images in the database, a verification rate of 96.5% was achieved at a False Acceptance Rate (FAR). The
experiments simulated both verification and identification systems and the results compared to those reported by state-of-
the-art works. In the identification scenario, rank-one accuracy was achieved. To the best of our knowledge, this is the
highest rank-one score ever achieved for the FRGC v2 database when compared to results published in the literature.
Keywords: 3D Face Recognition, Surface Interpenetration Measure (SIM), Range Image Registration.
I. INTRODUCTION
FACE recognition is a very challenging subject. So far,
studies in 2D face recognition have reached significant
development, but still bear limitations mostly due to pose
variation, illumination, and facial expression [13]. One way
to overcome such limitations is the use of face depth
information. In the 90’s, 3D face recognition stood out due
to advances of 3D imaging sensors. However, 3D images
also have limitations, such as the presence of noise and
difficult image acquisition [7]. A common approach for 3D
face recognition is the use of registration techniques to
perform range image matching. The Iterative Closest Point
(ICP) [5] algorithm, or one of its variants, is usually sought
to accomplish this task. The Mean Squared Error (MSE),
minimized during the convergence process, is then used to
compute the similarity between two face images [8], [9],
[14], [17]. This approach can also be employed with
deformation techniques to model facial expressions,
minimizing its effects on face recognition [17], [19], [20].
The performance of both 2D and 3D face recognition
systems can be verified on the Face Recognition Grand
Challenge (FRGC) [14]. The FRGC is an international
benchmarking composed of six challenging problems with
its expected performance guidelines. The third experiment,
i.e. ROC III, regards the 3D face recognition problem using
a challenging database composed of 4,007 images from 466
subjects displaying different facial expressions. For this
experiment, the goal is to achieve a verification rate of
98.0% at a False Acceptance Rate (FAR) of 0.1% [14].
Initially, the 3D face image can be acquired by different
techniques, such as laser and structured light scanners [15].
In the preprocessing stage, the input image is smoothed and
the face region is automatically detected and segmented.
Then, some facial feature points are detected to be used
during the matching process [11]. Each segmented face
region is matched with its corresponding one in the database.
A Simulated Annealing-based approach (SA) for range
image registration runs the process by using two robust
measures to assess alignment precision: (1) M-estimator
Sample Consensus (MSAC) [37], and (2) Surface
Interpenetration Measure (SIM) [11]. In previous works, the
SIM showed to be a discriminatory measure when working
on face images [4], [15].
This paper is organized as follows. Section II discusses
the main related works. Section III presents details about the
images available in the FRGC v2 database. Section IV
describes the preprocessing stage. Section V regards the