Iris localization using local histogram and other image statistics

Reliable iris localization using integral projection function and 2D-shape properties

“…The iris pattern remains stable over the entire life of a subject, except for some minor changes that occur in the early stages of life after birth [3,4] . A typical iris recognition system includes eye image capture, iris segmentation, feature extraction, and matching [2,3,5] . The term segmentation generally refers to the extraction of features from an image [6,7] .…”

“…Iris segmentation plays a critical role in the overall recognition accuracy because the accuracy of the other modules also depends on its performance. The iris segmentation module localizes the pupillary (iris inner boundary/pupil circle) and limbic (iris outer boundary/iris circle) boundaries and then detects and excludes any other superimposed noises, such as the eyelids, eyelashes, and/or specular reflections [1][2][3]5] . Researchers have proposed different methods for iris segmentation, including histogram-and thresholding-based techniques [2,5] , gradients and edge detectors [3,4] , and genetic algorithms and fuzzy logic [8,9] , among others.…”

“…[1,2,5]) because iris localization is considered to be the most challenging and time-consuming system module compared with other system modules. Daugman [3] used an integrodifferential operator (IDO) to localize the iris in an eye image.…”

“…Similarly, Wildes [4] used the gradient edge image and circular Hough transform (CHT) to localize the iris. Given that both the IDO and CHT localize the iris by using a circle approximation, the algorithms that use these operators generally take longer compared to their counterparts that use the histogram-and thresholdingbased techniques [1,2,5,10,11] . Khan et al [2,5] used histogram-, gradients-, and thresholding-based techniques to localize the pupil and iris circles.…”

“…Given that both the IDO and CHT localize the iris by using a circle approximation, the algorithms that use these operators generally take longer compared to their counterparts that use the histogram-and thresholdingbased techniques [1,2,5,10,11] . Khan et al [2,5] used histogram-, gradients-, and thresholding-based techniques to localize the pupil and iris circles. Khan et al [2] used eccentricity and Ibrahim et al [5] used the standard deviation of the position coordinates of pixels in an object to accurately find the pupil in a binary image.…”