A Dynamic and Multiresolution Model of Visual Attention and Its Application to Facial Landmark Detection

摘要

We describe a novel dynamic and multiresolution attention scheme for the generation of visual saccades and its application to locate candidate regions for facial feature recognition. The low-level, data-driven attention model suggested herein, employs a nonlinear sampling lattice of oriented Gaussian filters and uses small oscillatory movements to extract local image characteristics (conspicuity). As the sampling grid moves over the image, multiresolution ``evidences'' of local features are accumulated in a short-term visual memory. We propose a simple integration technique that computes the saliency surface iteratively across saccadic movements. Simulation results on face images demonstrate the applicability of our approach.