Comparison of Feature Extraction Technique for Segmentation in Human Iris Recognition Under Uncontrolled Environment Using CNN Algorithm with SVM Classifier
Busim Naga Siddu Karthik, G. Ramkumar
Abstract
The main aim of this study is to compare the radical segmentation of human iris using two different machine learning algorithms in an uncontrolled environment image dataset. Materials and method: Images taken from MMU iris dataset, Convolutional Neural Network (CNN) model, and Support Vector Machine (SVM) model implemented to segment iris in uncontrolled environment image with 50 samples per group. Results: MATLAB simulation result shows that CNN has accuracy of 94% and SVM has 72% in segmenting iris. Attained significant accuracy (0.001) in SPSS statistical analysis. Conclusion: For the given images, proposed CNN shows better accuracy than SVM classifier in iris segmentation tasks.
Topics & Concepts
Artificial intelligenceSupport vector machineComputer sciencePattern recognition (psychology)Iris recognitionConvolutional neural networkSegmentationIRIS (biosensor)Classifier (UML)MATLABFeature extractionBiometricsOperating systemBiometric Identification and Security