Feature Level Fusion of Biometric Images Using Modified Clonal Selection Algorithm

Abstract

In feature level fusion, biometric features must be combined such that each trait is combined so as to maintain feature-balance. To achieve this, Modified Clonal Selection Algorithm was employed for feature level fusion of Face, Iris and Fingerprints. Modified Clonal Selection Algorithm (MCSA) which is characterized by feature-balance maintenance capability and low computational complexity was developed and implemented for feature level fusion. The standard Tournament Selection Method (TSM) was modified by performing tournaments among neighbours rather than by random selection to reduce the between-group selection pressure associated with the standard TSM. Clonal Selection algorithm was formulated by incorporating the Modified Tournament Selection Method (MTSM) into its selection phase. Quantitative experimental results showed that the systems fused with MCSA has a higher recognition accuracy than those fused with CSA, also with a lower recognition time.