A novel GAAC optimisation algorithm for multimodal fusion score decision making in secured biometric systems

Abstract

Increased use of biometric systems on a global scale almost for all services have seen an increasing trend in research trying to improve the quality of authentication and containment of features extracted. A multimodal biometric system based on fusion score decision making has been proposed in this paper using a hybrid evolutionary framework. Genetic and ant colony optimisation (GAAC) algorithm has been presented and implemented on features of three biometric traits namely iris, fingerprint and finger vein to obtain a decision on the authenticity of the claiming individual. Features have been extracted using a frequency domain ridgelet transform as they are better able to approximate the fine component of ridges present on the fingerprint. The proposed hybrid technique is experimented on images from CASIA image database and efficiency metrics such as classification accuracy, positive find and negative find have been computed. The computational time has also been observed to be quite satisfactory due to fast converging nature of the hybrid combination.