Image quality assessment for iris biometric

Abstract

Iris recognition, the ability to recognize and distinguish individuals by their iris pattern, is the most reliable biometric in terms of recognition and identification performance. However, performance of these systems is affected by poor quality imaging. In this work, we extend previous research efforts on iris quality assessment by analyzing the effect of seven quality factors: defocus blur, motion blur, off-angle, occlusion, specular reflection, lighting, and pixel-counts on the performance of traditional iris recognition system. We have concluded that defocus blur, motion blur, and off-angle are the factors that affect recognition performance the most. We further designed a fully automated iris image quality evaluation block that operates in two steps. First each factor is estimated individually, then the second step involves fusing the estimated factors by using Dempster-Shafer theory approach to evidential reasoning. The designed block is tested on two datasets, CASIA 1.0 and a dataset collected at WVU. Considerable improvement in recognition performance is demonstrated when removing poor quality images evaluated by our quality metric. The upper bound on processing complexity required to evaluate quality of a single image is O(n 2 log n), that of a 2D-Fast Fourier Transform. This work wouldn't have been possible without the help, encouragement, tutelage, and guidance bestowed upon me over the past year of my Master's program. I would like to take this opportunity to thank those individuals for their time and patience. Firstly, I would like to express a deep heartfelt thanks to my advisor and committee chair, Dr. Natalia Schmid. She has been very patient with me and has provided me with invaluable guidance and encouragement, not only towards the completion of this work, but as a researcher. My graduate Committee members, Dr. Bojan Cukic and Dr. Larry Hornak deserve a heartfelt thanks as well, for their constructive criticisms and invaluable feedback.

Multi-angle sclera recognition system

“…For segmentation evaluat the segmentation accuracy of iris boundary For feature evaluation, we applied 19 by 19 window to downsample the image and che descriptors in the downsampled image. In used Dempster Shafer Theory to c segmentation, and feature scores together overall quality score [39,40].…”

Quality metrics influence on iris recognition systems performance

“…In order to express MOTION_BLUR, both the relative magnitude (strength) and the direction (angle) of the image motion have to be calculated. The computation method followed in this case is the one explained in [8]. Since there are two measurable components of motion blur, instead of considering the two different metrics separately, authors have decided to calculate a single value, given higher weight to magnitude, as considered more influential than direction.…”

An Automatic Algorithm for Evaluating the Precision of Iris Segmentation

“…The number of acquisitions per eye ranges from 2 to 17. This dataset has been identified as non-ideal [14], [12], with effects consisting of heavy occlusions and strong illumination variation. The WVU Off-Angle dataset consists of 560 images (early version), representing 140 different classes captured at a resolution of 720 x 576.…”

“…The other important research is image quality assessment [10], [11], [12], [13], [14], [15], [16]. Most of the referenced papers on nonideal iris recognition also propose a set of solutions to compensate or detect the effects that noticeably degrade iris images and thus influence recognition performance.…”

“…It has been previously reported that incorporation of iris image quality metrics into a matcher metric can improve recognition performance [14], [12]. However the quality metrics proposed in those works assume perfect iris image segmentation.…”