A box-counting fractal dimension for feature extraction in iris recognition

摘要

Iris recognition is one of a popular Biometrics technique for identifying people. Normally, iris color and texture have been created since the first three months and they will be completed within one year. Iris will then remain unchanged until the end of one's life. Both sides of the iris eyes are different. Moreover, the iris is unique and independent for genealogy even though in twins. Iris recognition has several sequence steps, pre-processing, features extractions, post-processing, and matching. In this paper, a box-counting fractal dimension and low-high pass filters have been used to extract iris features. This method can speed up a process by reducing a number of features. The experimental results show 91.22% matching accuracy with approximately 2.76 seconds for each matching.