Implementation and Testing

Having extensively discussed each module in our face detection, normalization and recognition system, we shall now describe an implementation of the complete system. We shall also briefly discuss the software design and the constraints on performance and efficiency. The system's user-friendly interface is then described. It facilitates the control of the algorithm and displays the localization and recognition results visually. To test the feature detection capabilities of the system, the localization results are verified for a few arbitrary input images. A large, standard database of faces is then used to test the recognition rates of the system. We follow the recognition and localization tests with a sensitivity analysis that provides us with some insight on the inter-dependence of these two components (recognition and localization) in the overall algorithm.