Estimation of perceived image blur

Sharpness is an important basic attribute of image quality. The spread of the blurring kernel is a good objective correlate of sharpness. Hence, by estimating the spread of the blurring kernel from a blurred image, lhe sharpness in the image can be predicted. Kayargadde and Martens (J 994) have presented an algorithm for estimating the spread of the blurring kernel from an image. In the algorithm, a global estimate is obtained by combining local estimates at prominent edge locations in the image. These local estimates at edges are arrived at by combining local derivatives. In this paper scaled unsharpness in images is correlated with a measure of sensory strength of blur that is obtained by using the blur estimation a1gorithm. We show that the estimates of sensory strength of blur obtained using the algorithm correlate well with the perceived unsharpness.