ant-CBIR: A New Method for Radial Furrow Extraction in Iris Biometric

Iris recognition has evolved from first to second generation of biometric systems which capable of recognizing unique iris features such as crypts, collarette and pigment blotches. However, there are still ongoing researches on finding the best way to search unique iris features since iris image contains high noise. The high noise iris images (noisy iris); usually give the biometric systems to deliver erroneous results, leading to categorizations where the actual user is labeled as an impostor. Therefore, this study focuses on a novel method, targeted at overcoming the aforementioned challenge. We present the use of ant colony based image retrieval (ant?CBIR) technique as a successful method in recognizing the radial furrow in noisy iris. This method simulates the behavior of artificial ants, searching for pixel values of radial furrow based on an optimum pixel range. The evaluation of accuracy performance with and without the ant-CBIR application is measured using GAR parameter on UBIRIS.v1. Results show that the GAR is 79.9% with ant-CBIR implementation. The implication of this study contributes to a new feature extraction that has the ability of human-aided computing. Moreover, ant-CBIR helps to provide cost effective, easy maintenance and exploration of a long term data collection.