A Learned Saliency Map for Eye Detection

摘要

In this paper, a method of constructing saliency map based on learning method is proposed. The saliency map is constructed by nonlinear combination of features, and the nonlinear function is approached through a feed forward artificial approached through a feed forward artificial neutral network (ANN) whose weight is trained by examples. The efficiency of the proposed method is verified in the eye detection problem. Experiment show that the learned saliency map may be efficiency used in the eye detection problem.