An improved model of saliency-based visual attention using Fuzzy Interface System

This study presents a novel method of generating saliency map in the basic saliency-based visual attention modeling proposed by Itti [1]. This can be applied in further processing such as object detection or recognition purpose. There are three conspicuity maps in saliency-based visual attention model which are derived from three different channels of color, intensity and orientation. The previous methods of generating saliency map are based on linear combination of these maps. Our proposed method employs Fuzzy Interface System (FIS) to combine three conspicuity maps instead of simple superposition which seems not reasonable biologically. Experimental results indicate the effectiveness of our proposed strategy in finding desired object with more accuracy in a datasets containing 32 images of emergency triangle with natural environment background. Furthermore, comparison with basic saliency model is provided to justify the capability of our proposed strategy.