A Genetic Algorithm based Method for Face Localization and pose Estimation

This paper introduces a new method using genetic algorithms (GA's) for face localization. Face images often have a background that can affect on the face localization algorithm. So, finding the exact location of face after localization algorithm is crucial. In our algorithm, input image is first enhanced by means of histogram equalization. Then connected components are determined by applying a region growing algorithm (coarse segmentation), followed by computing the fit ellipse for face area and at least exact location of face is found by genetic algorithms method. To check the utility of our proposed algorithm, we have define d tilt error and translation error that are the difference between obtained and real tilts and centers of localized face respectively. Simulation results on the ORL database show the effectiveness of this method. As, mentioned errors are obtained 4.2 degree and 5.4, respectively.