Iris Recognition Using Symmetric Graph Structure Based Pattern Matching

Automatic human identification using biometric information like Iris, leads on to the significant progress in the field of computer vision as it returns better authenticity and accuracy compared to other biometric recognition. This is because of its non-contact acquisition and user-friendly interface. But for unconstrained environment, it suffers when facial expression changes and light intensity differs. In this work, a novel approach for iris recognition called Multi Variant Symmetric Ternary Local Pattern (MVSTLP) is presented using the fundamental idea of pattern matching with the aim to find similarity between scene iris image and query iris pattern image by extracting distinctive features from them, where scene iris image is logically divided into number of query pattern size candidate windows. MVSTLP focuses on neighbour pixels selection in symmetric way within small image area and has unique ability to prioritize distinct feature extraction by establishing strong association between pixels. In effect of these, it can track very minute variations in image property and able to localize iris pattern within the scene iris image very accurately.