Facial features detection by saccadic exploration of the Gabor decomposition and Support Vector Machines
Fabrizio Smeraldi, N. Capdevielle, Josef Bigün · 1999
Facial features detection is of primary importance to face authentication and recognition. In this paper, we present an attention-driven approach to eyes and mouth detection inspired by the human saccadic system. The algorithm is centred around a log-polar retinotopic grid that is used to sample the Gabor decomposition of the image. Detection is achieved by displacing the grid according to a saccadic pattern. Saccade planning is performed using eye and mouth models implemented by means of Support Vector Machine classifiers.