Genetic Model Optimization for Hausdorff Distance-Based Face Localization

In our previous work we presented a model-based approach to perform robust, high-speed face localization based on the Hausdorff distance. A crucial step during the design of the system is the choice of an appropriate edge model that fits for a wide range of different human faces. In this paper we present an optimization approach that creates and successively improves such a model by means of genetic algorithms. To speed up the process and to prevent early saturation we use a special bootstrapping method on the sample set. Several initialization functions are tested and compared.