Analysis of webcam fingerprints for user authentication

Webcam fingerprint authentication is a new passwordless authentication method that aims to facilitate the user authentication process in order to make it more usable and still robust. This thesis analyzes the extraction of the webcam fingerprint, a noise pattern which identify each webcam among the others, and the process of fingerprint matching using some sets of photos acquired through an acquisition campaign. In the thesis there are explained the studies behind the construction of the capture script used in the campaign. Then, after creating the datasets with the different photos, there was the application of two different algorithms, the Fridrich algorithm and the ToothPic algorithm, for the fingerprint extraction and the fingerprint matching. Finally, the results obtained are showed by means of ROC curves. These ROC curves show the robustness of the webcam fingerprint authentication process analyzing the false positive rate and the true positive rate.