Vous êtes sur la page 1sur 6

HYBRID INTERPOLATION SUPER RESOLUTION BASED

ENHANCEMENT OF IRIS IMAGES

Hassan Aftab, Asif Butt, Umer Munir, Sheryar Malik, Omer Saeed
National University of Sciences and Technology, Pakistan
hassanaftab735@yahoo.co.uk

ABSTRACT
Iris recognition is a method of biometric authentication that uses pattern
recognition techniques based on high-resolution images of the iris. In this paper we
propose a method to obtain high resolution iris image from low resolution image
for facilitating the recognition process using hybrid interpolation super-resolution
technique that switches between New Covariance based interpolation and
Curvature based interpolation to produce sharp and refined iris images. The results
show the visually improved quality of Iris for recognition.

Keywords: Hybrid Interpolation, Super Resolution, Iris, Edges.

1 INTRODUCTION quality image for processing, thereby facilitating the


use of iris recognition process. The proposed method
Biometrics is the automated use of physiological is a fast hybrid approach for enhancement of iris
or behavioural characteristics to determine or verify images. The hybrid approach is an improvement of
identity. A distinction may be drawn between an method proposed by [5], [6]. In previous work, the
individual and an identity; the individual is singular, similar algorithm was employed on aerial images
but he may have more than one identity, for example [10]. The present work shows the effectiveness of the
ten registered fingerprints are viewed as ten different algorithm to obtain high resolution iris images for
identities. Iris’ are composed before birth and, except the iris recognition system. Moreover cropping
in the event of an injury to the eyeball, remain fused with hybrid image interpolation algorithm
unchanged throughout an individual’s lifetime. Iris- developed in Matlab gives the flexibility of
scan technology has been established as one of the extracting iris from the face of an individual for
biometrics that is very resistant to false matching and identification or surveillance. The results proved the
fraud. The false acceptance rate for iris recognition potential of the image for being utilized in iris
systems is 1 in 1.2 million, statistically better than recognition systems for enhancing e-security and
the average fingerprint recognition system. The real achieving maximum level of infallible security
benefit is in the false-rejection rate, a measure of measure. This paper is organized as follows: Section
authenticated users who are rejected [1]. For II presents the proposed algorithm. Section III
localizing the Iris and performing segmentation to presents the experimental results of our proposed
extract Iris pattern a high resolution iris image is technique. The paper is concluded in Section IV.
essential. Interpolation which is sometimes called re-
sampling is an imaging method to increase the 2 PROPOSED ALGORITHM
number of pixels in a digital image. Image
interpolation addresses the problem of generating a Edge detection is an important parameter in iris
high-resolution image from its low-resolution recognition system. The proposed algorithm
version. There are many methods for image differentiates between edge points and smooth areas
interpolation and commonly used ones in medical using the four neighbors of new interpolated points.
field include nearest neighbour, bilinear, bicubic and The difference of maximum and minimum values of
cubic-spline [2], [3]. all four points when compared with a pre-defined
The related work on enhancement of iris images threshold determines the presence of an edge or
is done by [4] which predicts the prior relation smooth area. If the difference exceeds the threshold
between iris feature information of different bands it is considered an edge otherwise the point belongs
and incorporates this prior into the process of iris to smooth area. Edges are determined using the new
image enhancement. The hybrid image interpolation covariance based method, whereas the smooth areas
super resolution algorithm developed in Matlab was are handled by the iterative curvature based approach.
employed on the iris’ image taken from Chinese The detail of the algorithm is explained in the
academy of sciences (CASIA). It yielded superior ensuing paragraphs.
2.1 Interpolation Scheme neighbouring low resolution pixels value, [a1, a2, a3,
a4, b1, b2, b3, b4, d1, d2, d3, d4, e1, e2, e3, e4] are
The interpolation scheme of proposed approach the four neighbouring pixels of low resolution pixels
is shown in Fig. 1. In this technique a low resolution [C1, C2, C3, C4] and [w, x, y, z] are the high
image is taken and padded with zeros. Next bilinear resolution coefficients. Using the concept of four
interpolation is performed at the four corners of equations and four unknowns the high resolution
padded image. Now the four diagonal neighbors of coefficients are estimated [6].
few points are available, therefore new data points
are calculated using these diagonal neighbors. In the (7)
end the remaining data points are determined using
four horizontal and vertical neighbors to produce a where ‘c5’ is the high resolution pixel value. Fig.
Super Resolved image. 2 shows the complete process of new covariance
based interpolation [6].
2.2 New Covariance Interpolation
2.3 Curvature Interpolation
New covariance based interpolation has been
used for interpolation in edges. Covariance is For interpolation in smooth areas curvature
employed by making a circular mask around every based method [7] has been employed. Curvature
unknown high resolution pixel and estimating the based interpolation performs bilinear interpolation
high resolution covariance coefficients from known along the direction where second derivative is lower.
low resolution covariance. The process is It is followed by iterative refinement of the
computationally complex as it keeps on looking for interpolated point following the isophote curve.
edges in that circular mask around the high Isophote curve is an intensity level curve which
resolution pixel until optimal MMSE (minimum compares present value with previous and next value
mean square error) is not achieved. The new and change the intensity of the present value. In this
covariance based interpolation employees the way the linear curve is changed into isophote curve
geometric duality property between the low- [8]. In case of diagonals, it finds the difference of
resolution covariance of four neighbour pixels and intensity levels between diagonals at opposite
the high resolution pixel. Geometric duality actually direction and performs bilinear interpolation, where
couples pair of pixels at different resolution but in the difference is less. The differences ‘V1’ and ‘V2’
same orientation. Since four near pixels are used we in two different directions are presented in [7], [8].
have the fourth order linear interpolation equation [6]
given as:

where ‘P’, ‘Q’, ‘P'’ and ‘Q'’ are intensity values


of the four neighbor points of interpolated pixel. The
where ‘Y’ is the required interpolated point in direction in which second derivative is lower,
high resolution image, ‘α’ is the high resolution bilinear interpolation is performed [7] as given in
covariance coefficients and ‘Y'’ is neighbouring equation 10 and 11.
pixels to interpolated point at low resolution image.
The neighbouring pixels of interpolated point are
known, so the only thing required is the high
resolution covariance coefficients [6].

(3)
where ‘P1’ and ‘P2’ are interpolated points in
two different directions. Next, iterative refinements
(4) are carried out on the interpolated points to follow
(5) the isophote curve. Fig. 3 shows the complete
(6) process of iterative curvature based interpolation [7].
where [C1, C2, C3, C4] are the four
Figure 1: Illustration Interpolation scheme of proposed method

Figure 2: New Covariance based Interpolation Scheme

Figure 3: Iterative Curvature based Interpolation Scheme


3 EXPERIMENTAL RESULTS In Fig. 5, an original iris image of size
320x280 is taken and small portion of this image of
The proposed algorithm is tested on both gray size 45x30 is cropped out of it as shown. This
scale and RGB iris images of CASIA database [9]. cropped image is then zoomed up using windows
In Fig. 4, a comparison of Peak Signal to Noise photo gallery and using new proposed method to
Ratio (PSNR) of ten interpolated RGB iris images size of 350x200. In Fig. 6, images of size 50×50
using nearest neighbour, bilinear, bicubic, iNEDI have been interpolated to size 400 x 400 using
and proposed method is plotted using MatLab. The various techniques such as zooming, nearest
PSNR values of different methods have been neighbour interpolation, bilinear interpolation,
summarized in tabular format in table 1. The bicubic interpolation, iNEDI method and proposed
average PSNR for each method is calculated which interpolation method. The result shows the visual
are 30.353 for new method, 30.138 for iNEDI quality improvements of new proposed method as
method, 19.463 for Bilinear method, 22.506 for compared to others mentioned. The result shows
Bicubic method and 17.669 for Nearest neighbour visually that employing the enhancement can help
method. The results clearly show the improved in identification of iris image pattern matching
PSNR performance of proposed method as compare techniques.
to conventional methods.

Figure 4: Matlab plot of PSNR of different interpolation methods of RGB iris images.

Figure 5: Cropping Application of New Proposed Method

Table 1: The table summarizes the PSNR in Decibels (dB) of Iris images enhanced using different interpolation
methods.

Images 1 2 3 4 5 6 7 8 9 10
Hybrid 30.37 29.9 30.83 30 29.93 30 30.3 30.4 31 30.8
iNEDI 30.27 29.84 30.44 29.8 29.43 29.9 30.1 30.13 30.9 30.57
Bilinear 19.53 18.96 19.87 19.35 19 19.33 19.41 19.58 19.95 19.65
Bicubic 22.56 22.11 22.93 22.11 22.45 22.41 22.55 22.59 22.5 22.85
Nearest 17.58 17.36 17.94 17.46 17.37 17.46 17.53 17.6 18.24 18.15
(a) Original image

(b) Simple zooming of image (e) Bicubic interpolation

(c) Nearest neighbor interpolation (f) iNEDI interpolation method

(d) Bilinear interpolation (g) Hybrid interpolation method

Figure 6: Interpolation methods employed to enhance 50×50 size iris image taken from CASIA database to
200×200 size for visual quality comparison. a) Original image, (b) zoomed image, (c) Nearest neighbor
interpolation, (d) Bilinear interpolation, (e) Bicubic interpolation, (f) iNEDI interpolation, (g) Hybrid
interpolation

4 CONCLUSION methods.

The proposed method is a hybrid technique for 5 REFERENCES


enhancing iris images in order to aid the
recognition process of iris recognition system. The [1] Nanavati S, T. M: Biometrics Identity
algorithm employs new covariance based Verification in A Networked World, New
interpolation for edges and iterative curvature based York: John Wiley & Sons, In (2002).
interpolation for smooth areas. A threshold is [2] Rafael C. Gonzalez, R. E: Interpolation
selected by performing an iterative experiment. Techniques, Digital Image Processing,
This threshold differentiates between an edge and Pearson Prentice Hall (2007).
smooth areas. The results showed improved PSNR [3] HS Hou, H.C: Cubic Splines for Image
performance and visually enhanced iris images with Interpolation and Digital Filtering, IEEE
the propose method as compare to conventional Transactions Acoustics, Speech, Signal
Processing. , pp. 508-517 (1978).
[4] Huang J. T.T. M.L. W.Y: Learning Based
Resolution Enhancement of Iris Images,
British Machine Vision Conference., pp. 153-
162 (2003).
[5] X. Li, M. T. Orchard: New edge-directed
interpolation. s.l. : IEEE Trans. on Image
Processing, Vols. 10, pp. 1521-1527 (October,
2001).
[6] Nicola Asuni, Andrea Giachetti: Accuracy
improvements and artifact removal in edge
based image interpolation. s.l. : Proc. 3rd Int.
Conf. Computer Vision Theory and
Applications, pp. 8 (2008).
[7] Andrea Giachetti, Nicola Asuni: Fast
Artifacts- Free Image Interpolation. Leeds,
UK : Proceedings of the British Machine
Vision Conference (BMVC), pp. 10
(September, 2008).
[8] B.S Morse, D.S: Isophate-Based Interpolation,
IEEE International Conference on Image
Processing , pp 227-231 (1998).
[9] Database: Center for Biometrics and Security
Research. Retrieved 2009, from Institute of
Automation chinese Academy of Sciences:
http://www.cbsr.ia.ac.cn/english/IrisDatabase.
asp
[10] Aftab, Hassan Mansoor, Atif Bin Asim,
Muhammad ‘A New single image
interpolation technique for super resolution’,
INMIC 2008, 23-24 Dec. Karachi , Pakistan
2008, page 592-596

Vous aimerez peut-être aussi