How to plot an average auc curve in a biometric authentication system?

I am working on face verification for 100 persons,and I need to calculate the verfication performance. The input to the Matlab function is the similarity scores and true labels, and the out is the Equal Error Rate (EER) and the area under the ROC curve (AUC). I used this function to verify each person separately. Thus, I obtained 100 EER values, one for each person. I used them to calculate the average EER.

My question is: How can I draw an average ROC curve, and How can I calculate an average AUC?

I can plot the ROC curve for each person separately.

[Ive J]

I'm baffled a bit here, because AFIAK, AUC is a property of the model fitted on a set of samples (here 100 individuals), so how can you even draw a ROC for each observation? How does it work?

[Hana Ahmed]

@Ive J Suppose that there are six samples for each person. Suppose further that one sample from each person is used for training and the remaining five for testing. I obtain the genuine scores by comparing the five test samples from person one by the training sample of person one. I obtain the imposter scores by comparing all test samples from other persons (5*99) to the training sample of person one. The obtained scores with their correct labels are fed to the function which is used to calculate the EER and plot ROC.

[Riham Abdelkebir]

i'm working on the field for a fingerprint recognition using matlab with 3 diffrent ways to calculate the score but i found an issue in ploting the FPR and TPR and the ROC however its works in normal way with just one score ( similarity measure )