Vector quantization and multi class support vector machines based fingerprint classification

Fingerprint recognition, one of the important biometric verification tools plays an important role in maintaining good security in any sort of systems. The fingerprint recognition in this paper deals with the extraction of features followed by codebook formation by Linde Buzo Gray Vector Quantization. The system also adds a sophisticated multi class Support Vector Machine classifier with a comparator technique which enhances the efficacy of the system. The classifier obtains an overall accuracy of 94% fingerprint verification with no rejects.