High resolution biologically inspired salient region detection

This paper presents a novel method for salient region detection. With the same biologically inspired logic as Itti et al., we use wavelet decomposition to obtain a ´centre´ scale, and filtered versions of the original image for the ´surround´ scale. Our method yields saliency map that is of the same resolution of the investigated image, with a detailed outline and high- light of the salient object. Previous methods are discussed in detail and comparisons are made as a validation process for our method.