No-Reference Image Quality Assessment for Sharpness Based on Line Spread Function

A no-reference quality evaluation algorithm is presented in this paper, in order to estimate the sharpness of images in practice. Firstly, useful information is obtained by extracting edge from original images and removing their complex texture region. Then line spread function curves are plotted along the gradient direction of each edge pixel, so that their full width values at half maximum can form the sharpness parameter congregation. Finally, the changes of image sharpness can be known by analyzing that congregation’s static characteristics. Experiments are carried out in three aspects: the relationship between images and their sharpness, the algorithm’s stability and anti-noise capability, the sharpness assessing ability for defocus images. The results show that this algorithm is effective in sharpness evaluation in the case with no reference. Furthermore, it has high stability and noise-resist capacity, indicating extensive value for engineering application.