Purchase this article with an account.
S. Murray, P. J. Bex; Blur Perception and Discrimination in Naturally-Contoured Images. Invest. Ophthalmol. Vis. Sci. 2010;51(13):1820.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Blur is an important measure of image quality and clinical visual function. The magnitude of image blur varies across space and time under natural viewing conditions owing to changes in pupil size and accommodation. We examine the perception of image blur in synthetic images composed of naturally-curved contours whose blur could be carefully specified. We aim to provide a bridge between the highly controlled 1D experimental stimuli and complex natural scenes for blur analysis.
Stimuli were produced in Matlab from low pass filtered random noise and possessed a ‘1/F’ amplitude spectrum. Image blur was manipulated by increasing or decreasing the slope of the amplitude spectrum, Gaussian low-pass filtering or filtering with a Sinc function, which introduces phase reversals that simulate qualities of optical blur. For each method and with forced choice paradigms, blur discrimination thresholds were measured at a range of blur levels and the perceived blur of differently-filtered images were cross-matched.
For slope-filtered images, blur discrimination thresholds for over-sharp images were extremely high and blur could not be matched with either Gaussian or Sinc filtered images, suggesting that directly manipulating image slope does not adequately simulate the perception of blur. For Gaussian and Sinc filtered images, blur discrimination thresholds were dipper-shaped and were well fit with a variance discrimination model involving internal visual blur followed by a simple Weber fraction. Blur matches between Gaussian and Sinc filtered images were not consistent with equating the relative energy across spatial frequencies, but were in good agreement with the location of peaks in the second spatial derivative of the edges’ luminance profiles and with the zero bounded regions determined with the MIRAGE model of edge detection.
The presence of high spatial frequencies in an image does not guarantee that the image will appear less blurred than a low-pass filtered image with a similar amplitude spectrum. Collectively, these results show that the relative phases of image components, as well as their relative amplitudes, determines the level of perceived blur.
This PDF is available to Subscribers Only