Iris recognition approach for identity verification with DWT and multiclass SVM

The iris has been proven to be one of the most stable and accurate biometrics. It has been widely used in recognition systems to determine the identity of the individual who attempts to access secured or restricted areas (e.g., airports, ATM, datacenters). An iris recognition (IR) technique for identity authentication/verification is proposed in this research. Iris image pre-processing, which includes iris segmentation, normalization, and enhancement, is followed by feature extraction, and matching. First, the iris image is segmented using the Hough Transform technique. The Daugman’s rubber sheet model is the used to normalize the segmented iris area. Then, using enhancing techniques (such as histogram equalization), Gabor wavelets and Discrete Wavelets Transform should be used to precisely extract the prominent characteristics. A multiclass Support Vector Machine (SVM) is used to assess the similarity of the images. The suggested method is evaluated using the IITD iris dataset, which is one of the most often used iris datasets. The benefit of the suggested method is that it reduces the number of features in each image to only 88. Experiments revealed that the proposed method was capable of collecting a moderate quantity of useful features and outperformed other methods. Furthermore, the proposed method's recognition accuracy was found to be 98.92% on tested data.