An Iris Characteristic Expression Algorighm Research Based on Statistical Method

Several algorithms on characteristic extraction and matching in the iris recognition system are researched after tradition algorithms are analysed according to the iris geometry and the physiological characteristic. A combination of Principal Component Analysis (PCA) method based on statistical and Linear Discriminant Analysis(FLD) method, PCA+FLD method, is put forward in the iris characteristic expression, which overcomes the traditional flaw that linear transform base vector is invariable. The recognition rate change curve of PCA and PCA+FLD at different iris match methods is created in the JLUBRIRIS database developed by us. By experiment, the best recognition rate and the best characteristic space are obtained in the specific environment for our JLUBR-IRIS iris database, and the recognition rate achieves 97.5%, which proves accuracy and validity of this method.