Objective smoothness metric for just noticeable blur estimation

An objective blur measure is crucial for a variety of image processing applications. Traditional researches concentrate on a model estimating the amount of spatial high frequency. However, human vision detects blurriness might be influenced by the texture of image contents. To address the important issue, this paper presents a new objective metric designed as both measuring the inherent smooth texture and predicting the difference induced by blur distortion. The approach is based on the wavelet characterization of Besov function spaces and just noticeable difference of human vision. Then, the just noticeable blur is estimated from the derived psychometric function using the proposed metric and subjective scores. Experimental comparisons with state-of-the-arts quality metrics for public databases show that the proposed metric achieved high correlation with the general objective quality or subjective metrics.