Experiments and Results
In this chapter, a collection of experiments and results are provided. There are two main sections of experiments: those designed to support the theory discussed in Chap. 4, and a sufficiently complete performance analysis. The first set of experiments study the uniqueness and robustness properties of the Simple K-Space approach. First, an experiment designed to demonstrate that any small change in the sketch results in a corresponding change in the model, indicate a local type of uniqueness (in general). Then, in order to study the robustness of SKS, the model is fixed and the test sketch varied in relatively small increments from the sketch that constructed the (constant) model to a completely different sketch. Analysis of the accumulators from this experiment reveals that initially there is strong peak, which diminishes as the test becomes inconsistent with the model. Thus, SKS directly incorporates tolerance into the methodology. The second set of experiments includes testing the performance on synthesized sketch-based password and two databases of hand-drawn sketches (one includes pressure information and the other does not).
KeywordsPerformance analysis FAR FRR EER
- 1.M. Martinez-Diaz, J. Fierrez, C. Martin-Diaz, and J. Ortega-Garcia. DooDB: A Graphical Password Database Containing Doodles and Pseudo-Signatures. 12th Int’l. Conf. on Frontiers in Handwriting Recognition, pages 339–344, 2010.Google Scholar
- 3.B. S. Riggan, W. E. Snyder, X. Wang, and J. Feng. A human factors study of graphical passwords using biometrics. Proc. of the 36th German Conf. on Pattern Recognition, 2014.Google Scholar