Security provision for biometric authentication systems using Enhanced Nelder Mead Algorithm

Abstract

Since, their are numerous advantages of biometrics-based authentication systems over traditional security systems based on knowledge, they are susceptible to attacks that can decrease their security significantly. We analyze these attacks in the multibiometric system. We propose an attack system that uses a hill climbing procedure to synthesize the targeted templates and evaluate its achievability with experimental results conducted on large databases. Hill climbing attack is nothing but security attack based on generating artificial data, after analyzing the output; updating such data, so as to improve the output. This is done repeatedly till output is desire output. So that, several actions can be utilized to decrease the probability of such attacks and their result are also presented. Some of the measures are uniform quantization techniques, non-uniform quantization techniques and many more. We are using uniform quantization, as quantization is the process of mapping a set of continuous pixel values into a finite numbers of possible values. The template division can be done on the basis of uniform quantized method which replicates the principle of uniform or linear quantizer has all the quantization levels uniformly distributed in the interval.