Multi Biometric Authentication using SVM and ANN Classifiers
T. Thamaraimanalan, S Pranavakumar, Lingeswaran Ra, Kabileshar Rm, M Sarankarthick · SSRN Electronic Journal · 2021
Data variability can be categorized based on the capture, analysis and treatment of biometric sample dataset. Initial interpretation of the data is necessary and this knowledge must be integrated into the identification system and an important feature of biometrics should be tested. This paper provides an examination of various biometric quality definitions and interpretations. In comparison to a bogus synthetic or reconstructed sample, the real existence of a legitimately valid phenotype is an essential issue for biological authentication that involves the creation of new and appropriate security mechanisms. The suggested solution introduces a modern fake identification programmed tool which can be used to identify various kinds of illegitimate access attempts in multiple biometric systems. It is also very difficult to use with 25 general image quality measurement features taken from an image, which makes it ideal for real time applications, to separate valid and impostor samples using the classification of artificial neuronal network. The study of actual biometric samples' general image quality provides extremely useful details which could be used extremely effectively to distinguish them from false properties. The experimental findings, obtained with publicly accessible fingerprint, iris, and 2D facial data, suggest that the proposed method is highly competitive with other cutting-edge approaches.