Abstract: Saliency-based visual attention model is widely used in recent years. Saliency-based visual attention allows for seemingly real-time performance by breaking down the complexity of scene understanding into a fast temporal sequence of localized pattern recognition problems, in a computationally efficient manner. Based on Itti’s visual attention model, this paper evaluates the effects of specific parameters on the performance of saliency map. Statistical methods are applied to choose optimal parameters, which allowed to improve the performance of saliency map. Some experimental results show that the method can quickly and accurately detect targets in a complex background.

Key words: saliency map, parameter tuning, visual attention