Image quality evaluation based on saliency and feature similarity

Abstract

A full reference image quality evaluation method based on image saliency and feature similarity is proposed in this paper. The method first calculates the feature similarity map between the degraded image and the reference image. Then in order to emphasize the crucial role of the salient region in the image quality evaluation saliency map of reference image is extracted as weight of the feature similarity map. The saliency map is extracted by the method based on the combination of spatial domain and frequency domain. Finally, the degraded image quality score is predicted with the saliency map and the feature similarity map. The obtained scatter plots and nonlinear regression functions show that the scores of various types of degraded images predicted by this method are consistent with the subjective values.