Abstract: Existing computation methods of Human visual attention are rather impractical due to either the involved high computational complexity or the difficulties of selecting suitable parameters.In this paper,we propose a fast method to simulate attention selection in frequency domain.We set the amplitude spectrum to a constant and preserve phase spectrum,then transform them back into the spatial domain.After filtering,the resultant saliency map is quite similar to those of existing methods.In addition,we treat the salient information as weights to adjust several main existing criteria of image quality assessment.Experimental results show that our image quality assessment with attention selection is more consistent with human's subjective evaluation,besides its computational efficiency.