A Study on Pupil and Iris Segmentation of the Anterior Segment of the Eye

Ho-Chul Kang; Kwang Gi Kim; Whi-Vin Oh; Jeong-Min Hwang

doi:10.4258/jksmi.2009.15.2.227

Journal List > J Korean Soc Med Inform > v.15(2) > 1035526

Go to TopGo to Top Go to BottomGo to Bottom

TOOLS

Kang, Kim, Oh, and Hwang: A Study on Pupil and Iris Segmentation of the Anterior Segment of the Eye

Original Article

Journal of Korean Society of Medical Informatics

Journal of Korean Society of Medical Informatics 2009; 15(2): 227-234.

Published online: 30 June 2009

DOI: https://doi.org/10.4258/jksmi.2009.15.2.227

A Study on Pupil and Iris Segmentation of the Anterior Segment of the Eye

Ho-Chul Kang¹, Kwang Gi Kim¹, Whi-Vin Oh¹, Jeong-Min Hwang^2,³

¹Biomedical Engineering Branch, Division of Cancer, National Cancer Center, Korea.

²Department of Ophthalmology, Seoul National University College of Medicine, Korea.

³Seoul National University Bundang Hospital, Korea.

Corresponding Author: Kwang Gi Kim, Biomedical Engineering Branch, Division of Cancer, National Cancer Center, 111, Jeongbalsan-ro, Madu 1-dong, Ilsandong-gu, Goyang 410-769, Korea. Tel: +82-31-920-2241, Fax: +82-31-920-2242, kimkg@ncc.re.kr

Received 18 August 2008 Accepted 3 June 2009

Abstract

Objective

The goal of this study was to develop a novel pupil and iris segmentation algorithm. We evaluated segmentation performance based on a fractal model. Two methods were compared: Daugman's and our new proposed method.

Methods

We received 200 anterior segment images with 3,872×2,592 pixels. Here we present an active contour model that accurately detects pupil boundaries in order to improve the performance of segmentation systems. We propose a method that uses iris segmentation based on a fractal model. We compared the performance of Daugman's method and the proposed new method and statistically analyzed the results.

Results

We manually compared segmentation with the Daugman's method and the new proposed method. The findings showed that the proposed segmentation accuracy was about 2.5 percent higher than Daugman's method. There was a significant difference (p<0.05) between the under and over data between the two methods.

Conclusion

The results of this study show that the new proposed method was more accurate than the conventional method for the measurement of segmentation of the eye by CAD (Computer-aided Diagnosis).

Keywords: Pupil, Iris, Segmentation, Computer-aided Diagnosis, Image Processing

Figures and Tables

Figure 1

Flowchart of proposed method

Figure 2

(a) Original image, (b) Image of hue, (c) Image of saturation, (d) Image of value

Figure 3

Pupil segmentation result

Figure 4

Iris segmentation based on radiuses directions

Figure 5

The result of iris segmentation of fractal dimension method

Figure 6

The ratio of over segmentation

Figure 7

The comparison between Daugman's method and proposed method

Figure 8

The ratio of under segmentation

Figure 9

The comparison between Daugman's method and proposed method

Table 1

The result of statistical analysis of over segmentation

Table 2

The result of statistical analysis of under segmentation

Table 3

The comparison between Daugman's method and proposed method

Notes

This work was supported by a research grant from National Cancer Center, Korea (grant 0810122) from the Korea Science and Engineering Foundation (R01-2005-000-10875-0)

References

1. Lim JI, LaBree L, Nichols T, Cardenas I. A comparison of digital nonmydriatic fundus imaging with standard 35-millimeter slides for diabetic retinopathy. Ophthalmology. 2000. 107(5):866–870.

2. Yen KG, Hess D, Burke B, Johnson RA, Feuer WJ, Flynn JT. The optimum time to employ telephotoscreening to detect retinopathy of prematurity. Trans Am Ophthalmol Soc. 2000. 98:145–150. discussion 150-151.

3. van Norren D, van de Kraats J. Retinal densitometer with the size of a fundus camera. Vision Res. 1989. 29(3):369–374.

4. Tuulonen A, Airaksinen PJ, Montagna A, Nieminen H. Screening for glaucoma with a non-mydriatic fundus camera. Acta Ophthalmol (Copenh). 1990. 68(4):445–449.

5. Zeimer R, Zou S, Meeder T, Quinn K, Vitale S. A fundus camera dedicated to the screening of diabetic retinopathy in the primary-care physician's office. Invest Ophthalmol Vis Sci. 2002. 43(5):1581–1587.

6. Gili Manzanaro P, Carrasco Font C, Martin Rodrigo JC, Yanguela Rodilla J, Arias Puente A. Digital analysis of the optic disc with fundus camera: a study of variability. Arch Soc Esp Oftalmol. 2004. 79(3):125–130.

7. Schmitz-Valckenberg S, Fleckenstein M, Gobel AP, Sehmi K, Fitzke FW, Holz FG, et al. Evaluation of autofluorescence imaging with the scanning laser ophthalmoscope and the fundus camera in age-related geographic atrophy. Am J Ophthalmol. 2008. 146(2):183–192.

8. Daugman J. High confidence Recognition of Persons by Rapid Analysis of Iris Texture, IEEE Conference Publication. 1995. 05. 244–251. No. 408.

9. Daugman J. New methods in iris recognition. IEEE Trans Syst Man Cybern B Cybern. 2007. 37(5):1167–1175.

10. Huang H, Hu G. Iris recognition based on adjustable scale wavelet transform. Conf Proc IEEE Eng Med Biol Soc. 2005. 7:7533–7536.

11. Monro DM, Rakshit S, Zhang D. DCT-based iris recognition. IEEE Trans Pattern Anal Mach Intell. 2007. 29(4):586–595.

12. Klemencic A, Kovacic S, Pernus F. Automated segmentation of muscle fiber images using active contour models. Cytometry. 1998. 32(4):317–326.

13. Bogdanova I, Bresson X, Thiran JP, Vandergheynst P. Scale space analysis and active contours for omnidirectional images. IEEE Trans Image Process. 2007. 16(7):1888–1901.

14. Parker JR. Algorithms for image processing and computer vision. 1997. WILEY COMPUTER PUBLISHING, 1997.

15. Gonzalez RC. Digital image processing using MatLab. 2004. Seoul: ITC..

16. Otsu N. A threshold selection method from gray_level histograms. IEEE Trans Systems Man and Cybernetics. 1979. 79:62–66.

TOOLS

Similar articles