A Soft Computing Technique for Improving the Fidelity of Thumbprints Based Identification Systems

With the advent of new thumbprint identification techniques, accurate personal identification is now easy and cheaper with approximately zero false acceptance rates. This paper focuses on developing an advance feature for thumbprint based identification systems with the help of soft computing and 2D transformation which makes the technique more flexible and Fidel. The thumbprint images of individuals were scanned with the help of H3 T&A terminal for collecting self generated datasets. The thumbprints of self generated and standard datasets were trained to form a refined set which includes linear and angular displacements of thumbprint images. The new obtained features of refined datasets were stored in the database for further identification. In the proposed technique, the minutiae coordinates and orientation angles of the thumbprint of a person to be identified are computed and merged together for comparison. The minutia coordinates and orientation angles of a person are compared with the minutiae trained set values stored in the database at different linear and angular rotations for identity verification. The proposed technique was tested on fifty persons self generated and standard datasets of FVC2002, FVC2004 and CASIA databases. In the experimentation and result analysis we observed that the proposed technique accurately identifies a person on the basis of minutiae features of a thumbprint with low FNMR (False Non-Match Rate) values.