Feature Encoding and Selection for Iris Recognition Based on Variable Length Black Hole Optimization

Iris recognition as a biometric identification method is one of the most reliable human methods. It exploits distinctive pattern iris area. Typically, several steps are performed for recognition, namely, pre-processing, segmentation, normalization, extraction, coding and classification. In this article, we present novel algorithm that includes in addition to features extraction step feature selection. Furthermore, it enables selecting variable length by adapting our recent black hole optimization (VLBHO). first selection recognition. Our proposed segments-based decomposition according their relevance which makes more efficient terms both memory computation promising convergence. For classification, article uses famous support vector machine (SVM) Logistic model. The has been evaluated based on two datasets, IITD CASIA. finding optimizing encoding VLBHO superior benchmarks with an improvement percentage 0.21%.