In training phase of the proposed iris recognition system, for each iris k, DHMM (Discrete HMM), [[theta].sub.k] has been built .
For measuring the accuracy of individual left iris and right iris based unimodal recognition system, the critical paramerter, that is, the number of hidden states of DHMM, can affect the performance of the system.
In feature level fusion, optimum value of the number of hidden states of DHMM has been chosen and Figure 8 shows the comparison results with ROC curve.
Left-right iris likelihood ratio based score fusion method is a score fusion technique where the reliability of each modality is measured by using the output of DHMM classifier for both the left and right iris features.
To calculate the score fusion result, the DHMM outputs of individual iris (i.e., left and right) recognition are combined by a weighted sum rule to produce the final score fusion result.