Content area
Full Text
Abstract. Biometrics based identity-verification is regarded as an effective method for automatically recognizing, with a high confidence, a person's identity. This paper presents a new biometric approach to identity verification based on the tongue-print technology by means of a novel feature extraction method since tongue can be stuck out of mouth for inspection, and it is otherwise well protected in the mouth and is difficult to forge; this organ could be considered as a biometric factor. The tongue presents both geometric shape information and physiological texture information which are potentially useful in personal identification applications. Despite these obvious advantages for biometrics, little work has hitherto been done on this topic; nevertheless, the feasibility of tongue biometrics has been proved. Our tongueprint identification system employs tongueprint images and by means of feature extraction from wavelet coefficients achieves effective personal identification. The suggested system consists of two parts: preprocessing of acquired tongue images, and a feature extraction method to achieve each person's unique parameters. In order to have an experimental image database, digital pictures were acquired from 10 different people's tongues in prespecified time intervals. These people were chosen randomly both in gender and age although specifically from different age groups. Resolution of images and their acquisition conditions were uniform in all cases. By clustering different people's data and specifying the cluster centre, any new dataset would be compared to the claimed centre and his/her identity could be verified. Comparison of different mother wavelet transforms' results demonstrate the optimum framework. The experimental results have been verified at the end. The results from suggested procedure show that using texture factor for the tongue verification produces an efficient and reliable result. When the FAR is equal to 3.1%, we get the Genuine Accept Rate of 70%. These results once again demonstrate that the tongue biometric is feasible. Additionally it was observed that at the threshold of 182, the suggested procedure would have equal error rate. Final verification results depict close match with previous studies. Furthermore it was shown that application of tongue's texture - solely as a biometric factor considering possible interfering of tongue shape complications resulted by its probable and perhaps voluntary changes -could be regarded certainly efficient and even more reliable with high recognition rate.