• No results found

Dorsal Hand Vein Enhancement and Fake Vein Detection

N/A
N/A
Protected

Academic year: 2020

Share "Dorsal Hand Vein Enhancement and Fake Vein Detection"

Copied!
5
0
0

Loading.... (view fulltext now)

Full text

(1)

2017 2nd International Conference on Artificial Intelligence: Techniques and Applications (AITA 2017) ISBN: 978-1-60595-491-2

Dorsal Hand Vein Enhancement and Fake Vein Detection

Shou-kun JIANG, Fu LIU

*

, Bing KANG, Zi-yue YOU and Yu-xuan ZONG

College of Communication Engineering, Jilin University, Changchun 130025, China

*Corresponding author

Keywords: Dorsal hand vein, Guided filter, Image enhancement, Fake vein detection.

Abstract. In order to extract better vein feathers, image preprocessing is necessary. So this paper propose a new approach to enhance images. The enhancement algorithm uses guided filter (GF) to process hand vein images. The guided filter is used as an edge-preserving smoothing operator. The guided filter enhancement algorithm is effective comparing with bilateral filter (BF), histogram equalization (HE), adaptive histogram equalization algorithm (AHE) and contrast limited adaptive histogram equalization (CLAHE). We use several methods to enhance dorsal hand vein images, the recognition rate with guided filter is the best. For the security, a fake vein detection algorithm is used to discriminate the real vein and fake vein images.

Introduction

With the rapid development of society, biometric identification technology are also developing quickly. Hand vein recognition, as the important branch of biometrics, have gotten the attention of many researchers. As with other biometrics, hand vein also can offer high accuracy. Moreover, hand vein can ensure liveness [1] and can't be seen with naked eyes [2], which are advantages because these make vein pattern difficult be forged. Contrast to fingerprint and palm print feathers extraction, the disturbances of humidity and the state of skin for hand veins are very little [3]. So vein recognition have a remarkable ability to adapt to changing environments comparing with other biometric identification.

Hand veins exist under the skin, so we need some instruments to find the vein, one of them is infrared camera. The near infrared cameras are usually used to capture hand veins. To reduce the computation time and improve the calculation speed, the hand vein images are always low quality, so most of methods are preceded by a preprocessing step which can make it easier to extract vein features. Redhouane applied a double adaptive histogram equalization to enhance the contrast of the images [4]; Chanthamongkol used the contrast-limited adaptive histogram equalization in the enhancement of dorsal vein [5]; Guo Dan et al. used multi-scale vessel enhancement filtering to enhance the noise-depressed images [6]; Trabelsi used contrast limited adaptive histogram equalization, moreover histogram equalization and adaptive histogram equalization algorithm are used to contrast with[7,8]; Maurício Ramalho used median and wiener filter [9]; Ramsoful used median and Gaussian filter [10]; Wei used Gaussian and high pass filter [11]; Kumar used Mexican Hat to enhance image. With a view to the importance of pre-processing, the guided filter [12] is proposed to enhance the hand vein images. However, for the security, the fake vein detection is also important [13, 14], the paper also proposes a fake vein image detection algorithm.

The rest of the paper is organized as follows. In Section II and III, we present the contrast of several enhancement methods and recognition rate. In Section IV, we present fake vein detection. Finally Section V will make conclusion about our works.

Contrast of Several Enhancement Methods

(2)

(

)

:

1

k

i k i k i i i

k i

q

a p

b

a p

b

ω

ω

=

+

=

+

(1)

Here,

1

i

i k k

a

a

ω

ω

=

and

1

i

i k k

b

b

ω

ω

=

, k is the center pixel index of

ω

k, i is involved in

all the windows

ω

kthat contain i, p is input image, and q is output image.

In order to improve the contrast of the images, another processing has been done as fomulla (2).

(

)

*

,

min

en

a

q

q

m

p

q

q m

abs

p

q

=

+

=

(2) [image:2.612.105.508.250.380.2]

Where

q

en is the enhanced images and a is a matrix whose elements are 255 and size is same to p.

Figure 1. The source Images.

[image:2.612.173.457.477.548.2]

Figure 1 are the source images (left hand and right hand), and each size is 320x240. There are 380 people to be captured hand vein images. The maximum inscribed circle of hand image contour is used to search the maximum ROI. Then we rotate the circular region according to the midpoint of wrist. Finally a maximum square of the circular region is extracted and normalized. In Figure 2, the circle is the maximum inscribed circle, the pentagram is the center of the circle. Figure 3 is the circle ROI, Figure 4 is the square ROI after size normalization(128x128).

Figure 2. Vein image with circle. Figure 3. Circle ROI. Figure 4. Square ROI.

Figure 5-8 are the enhanced ROI with HE, CLAHE, BF and GF. HE can improve the contrast of image, however this algorithm can amply noise and enhance the contrast excessively(Figure 5); the bilateral filter(Figure 7), similar to HE, can enhance the contrast excessively; compared to HE, CLAHE(Figure 6) limits the degree of amplification and noise, but the degree of amplification is smaller than guided filter; guided filter(Figure 8) has the edge-preserving smoothing property, what's more, this algorithm can preserve and amply the gradient of edge, which makes the enhancement performance better than other algorithms.

(3)

Contrast of Recognition Rate

In this paper, the recogniton rate is used to test the effect of enhancement methods. We contrast HE, BF, CLAHE and GF, the result is listed in table 1. From the table, the recognition rate with GF is better than others, which can explain this enhancement algorithm is effective from the side.

Table 1. Contrast of recognition rate using enhancement methods. ERR(%)(right) ERR(%)(left)

HE 5.3 6.5

BF 3.8 4.7

CLAHE 2.1 3.6

GF 1.6 2.7

Fake Vein Detection

With the development of vein recognition, one critical issue is the use of fake vein images to carry out system attacks [14]. To overcome this problem, we captured some fake vein image which are paper printed, wearing disposable gloves and proposed a fake vein image detection algorithm. Figure 9 is a real vein image, and Figure 10 is a fake vein image.

[image:3.612.201.410.300.376.2]

Figure 9. Real vein image. Figure 10. Fake vein image.

In order to discriminate the real vein and fake vein, we use the contrast ratio. The fomulla is as (3):

(

)

2

(

)

,

=

,

*

,

i j

C

δ

i j

p i j

(3)

Where

δ

(

i j

,

)

=

i j

-

,

p i j

(

,

)

is the probability when the difference is

δ

betweed adjacent [image:3.612.188.420.484.660.2]

pixels and center pixel.

Figure 11. Contrast ratio between real vein and fake vein images.

(4)

Conclusion and Future Works

We have presented a new method for hand vein image enhancing. Contrast to other methods, guided filter enhancement can preserve the gradientof edge, and amply the degree of gradient which is important to extract hand vein feathers. We also proposed a fake vein image detection algorithm. The experimental results is good, but in the experiments exist some problems:

(1) Rotation of image using wrist midpoint: for the instability of wrist midpoint may result in the mismatching for the same hand vein images.

(2) Database: there are only 180 fake vein images, which is small contrast to real vein images, so the database need to expand. What’s more, there are some people whose hand veins are hard to be capture, for fat hands.

(3) Fake vein: only one kind of vein forgery is used to test the fake vein detection algorithm.

Acknowledgement

This research was financially supported by project: palmprint diagnosis acquisition system and industrialization based on IOT (2017C032-4).

References

[1] Ajay Kumar, and K. Venkata Prathyusha. Personal Authentication Using Hand Vein Triangulation and Knuckle Shape[J]. IEEE Transactions on Image Processing, 2009,18(9): 2127-2136.

[2] Maleika Heenaye- Mamode Khan, and Naushad Ali Mamode Khan. A New Method to Extract Dorsal Hand Vein Pattern using Quadratic Inference Function[J]. International Journal of Computer Science and Information Security, 2009, 6(3): 26-30.

[3] Liukui Chen, Hong Zheng, Li Li, Peng Xie and Shou Liu. Near-infrared Dorsal Hand Vein Image Segmentation by Local Thresholding Using Grayscale Morphology[C]. The 1st International Conference on Bioinformatics and Biomedical Engineering, Wuhan, 2007: 868-871.

[4] Lefki Redhouane, Benziane Sarah, and Benyetton Abdelkader. Dorsal hand vein pattern feature extraction with wavelet transforms[C]. International Symposium on Networks, Computers and Communications, Hammamet, 2014: 1-5.

[5] Sathaporn Chanthamongkol, Boonchana Purahong, and Attasit Lasakul. Dorsal Hand Vein Image Enhancement for Improve Recognition Rate Based on SIFT Keypoint Matching[C]. The 2nd International Symposium on Computer, Communication, Control and Automation, Singapore, 2013:174-177.

[6] Guo Dan, Zhexiao Guo, Huijun Ding, and Yongjin Zhou. Enhancement of Dorsal Hand Vein Image with a Low-Cost Binocular Vein Viewer System[J]. Journal of Medical Imaging and Health Informatics. 2015, 5(2): 1-7.

[7] Randa Boukhris Trabelsi, Alima Damak Masmoudi, and Dorra Sellami Masmoudi. A new multimodal biometric system based on finger vein and hand vein recognition[J]. International Journal of Engineering and Technology. 2013, 5(4): 3175-3183.

[8] Randa Boukhris Trabelsi, Alima Damak Masmoudi, and Dorra Sellami Masmoudi. A Novel Biometric System Based Hand Vein Recognition[J]. Journal of Testing and Evaluation, 2014, 42(4): 1-10.

(5)

[10] Pooja Ramsoful, and Maleika Heenaye-Mamode Khan. Feature Extraction Techniques for Dorsal Hand Vein Pattern[J]. Journal of Intelligent Computing. 2013, 4(3): 115-122.

[11] Shangqing Wei, and Xiaodong Gu. A Method for Hand Vein Recognition Based on Curvelet Transform Phase Feature[C]. International Conference on Transportation, Mechanical, and Electrical Engineering, Changchun. 2011: 1693 - 1696.

[12] Kaiming He, Jian Sun, and Xiaoow Tang. Guided Image Filtering[C], European Conference on Computer Vision, Heraklion. 2010: 1-14.

[13] Wang, Y. D., Zhang, D.,Qi, Q. Liveness detection for dorsal hand vein recognition[J]. Personal and Ubiquitous Computing, 2016, 20:447-455.

Figure

Figure 1. The source Images.
Figure 11. Contrast ratio between real vein and fake vein images.

References

Related documents

Press and hold the space bar and set the power switch to I (on) and continue holding the space bar until the following display appears.. Note: If the above display does

Also, a non-TCP congestion control algorithm was implemented on the router to optimize the retransmission timer and make the acknowledgement of request to be faster in order to

Ethernet passive optical networks (EPONs), which characterize the convergence of small-price Ethernet equipment and small-price fiber communications, appear to be

The views expressed by the students emphasize the role that mentors play in Initial Teacher Education as well as the importance attached to teaching practice

In-situ Scanning Electron Microscopy for Electron-beam Lithography and In-situ One Dimensional Nano Materials Characterization..

WORLD JOURNAL OF SURGICAL ONCOLOGY Kwak et al World Journal of Surgical Oncology 2013, 11 77 http //www wjso com/content/11/1/77 RESEARCH Open Access Breast cancer after

In addition, F values in wood properties remained relatively higher from the 1st to 25th annual ring from the pith, although F value in ARW rapidly decreased with each increase