Approach for image segmentation based on improved visual attention mechanism

The computational model of visual attention mechanism could be used to solve the deficiency of classical image segmentation algorithm such as setting parameter. In view of the Itti model has the problems that edge blur, low efficiency and the saliency map is not obvious, an improved visual attention mechanism is proposed in this paper. It introduces the edge enhancement equation, uses the combination strategy based on weight, and so on. The improved visual attention mechanism combines with the level set method. The binarization saliency map is evolved to obtain the image segmentation result. The experiments show that the proposed algorithm outperforms the SVM algorithm, the adaptive threshold algorithm and the Kmeans clustering algorithm in image segmentation, and it shows strong robustness.