A iris recognition algorithm based on ICA and SOM neural network

For overcoming defect of vector-based nondynamic and parameter specifies method which based on linear transformation feature extraction, a iris feature extraction method which based on the Independent Component Analysis (ICA) is advanced. This method almost removed redundancy of feature space and overcome the defect of traditional linear transformation feature-based vectors non-dynamic. And then Self-Organizing Maps(SOM) neural network is used for iris classification and recognition. As the experimental result shown that the recognition rate of three sample is 98.81%, 96.67% and 100%, respectively. The correctness and validity of this algorithm is proved by these experimental result.