International Journal on Advanced Science, Engineering and Information Technology, Vol. 7 (2017) No. 4-2: Special Issue on the Emerging Trends in Software Engineering and Soft Computing Applications, pages: 1510-1515, Chief Editor: Shahreen Kasim Editorial Boards: Rohayanti Hassan, Hairulnizam Mahdin, Mohd Farhan Md Fudzee & Azizul Azhar Ramli, DOI:10.18517/ijaseit.7.4-2.3392

Image Enhancement Technique at Different Distance for Iris Recognition

Rohayanti Hassan, Shahreen Kasim, Wan Ain Zubaidah Wan Chek Jafery, Zuraini Ali Shah

Abstract

Capturing eye images within visible wavelength illumination in non-cooperative environment lead to the low quality of eye images. Thus, this study is motivated to investigate the effectiveness of image enhancement technique that able to solve the abovementioned issue. A comparative study has been conducted in which three image enhancement techniques namely Histogram Equalization (HE), Adaptive Histogram Equalization (AHE) and Contrast Limited Adaptive Histogram Equalization (CLAHE) were evaluated and analysed. UBIRIS.v2 eye image database was used as a dataset to evaluate those techniques. Moreover, each of enhancement techniques were tested against different distance of eye image captured. Results were compared in term of image interpretation by using Peak-Signal Noise Ratio (PSNR), Absolute Mean Brightness Error (AMBE) and Mean Absolute Error (MAE). The effectiveness of the enhancement techniques on different distance of image captured was evaluated using the False Acceptance Rate (FAR) and False Rejection Rate (FRR). As a result, CLAHE has proven to be the most reliable technique in enhancing the eye image which improved the localization accuracy by 7%. In addition, the results showed that by implementing CLAHE technique at four meter distance was an ideal distance to capture eye images in non-cooperative environment where it provides high recognition accuracy, 74%.

Keywords:

iris recognition; image enhancement; histogram equalization; CLAHE

Viewed: 577 times (since abstract online)

cite this paper     download