The Elimination Eyelash Iris Recognition Based on Local Median Frequency Gabor Filters

The orientation of iris texture will influence the iris recognition rate and the 2 dimensional Gabor filters (2-D Gabor) has better selectivity in orientation and frequency, so the multi-scales and multi directions 2-D Gabor filters can be used to extract the iris features. But there are many shortcomings such as long extraction feature time and high feature dimensions to use the traditional 2-D Gabor filters. So the elimination eyelash influence iris recognition based on median frequency Gabor filters is proposed in this paper. Firstly, the iris is preprocessed by iris and eyelash localization, normalization and elimination eyelash shelter. Secondly, 2-D Gabor filters with 3 scales and 8 directions are constructed to extract the feature of iris after preprocessing. Then using the Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) further extract the feature and reduce the feature dimension. The experiment shows that the feature dimension is greatly reduced and the iris texture orientation is well considered by the proposed method compared with the traditional 2-D Gabor filters. The method in the paper can recognize the iris accurately and rapidly.