Decision level biometric fusion using Ant Colony Optimization

This paper presents a decision level fusion scheme for palmprint and hand vein biometrics using an evolutionary technique such as Ant Colony Optimization (ACO) to compute the fusion parameters by selecting them dynamically. A digital camera based imaging set up is employed for acquiring the palmprint images while Infrared thermal imaging is used for hand vein acquisition. A modified block limited phase only correlation (MBLPOC) function is utilized for representing the similarity between the acquired palm images and Gabor wavelets based features are extracted from the vein images. Since both the biometrics are represented in different domain, their integration has been done on the decision level by correctly choosing the decision thresholds; which can optimize the decisions taken by the individual matchers using ACO algorithm. The experimental results carried out on the database of 150 users are promising and thus confirming the usefulness of the proposed fusion system.