Abstract
Purpose: :
To evaluate whether vision is adapted to the orientation of the natural blur produced by the High Order Aberrations (HOA) of the eye.
Methods: :
Judgments of perceived blur were measured in 4 subjects in two 2AFC psychophysical tasks: (1) Pattern Preference experiment (10 natural images, 100 pairs, 2 choices) and (2) Image Classification-based experiment (1 face, 500 pairs, 6 graded choices). The optical blur was generated by convolution with the PSF estimated from 100 real different HOA patterns. Tilts, astigmatism and defocus were set to zero. The HOA were scaled to produce the same blur level, matching the subject’s strehl ratio, with different orientations. Pairs of images, displayed sequentially, were viewed monocularly under full Adaptive Optics correction of the subject’s aberrations. Observers selected the image that appeared sharper. In experiment 1, each pair was composed of one image always blurred with a constant HOA pattern whereas the other was blurred with a randomly selected pattern. 11 different constant patterns were tested including the subjects’ own and a 90º rotated version. Results were analyzed in terms of percentage of images blurred with the constant HOA judged as sharper. In Experiment 2, both images were blurred with different HOA patterns randomly selected. The images judged as sharper were identified as positive, the others as negative responses. In addition, graded responses were weighted to perceptually rank the HOA patterns. Correlations between the subject’s PSF and the PSF corresponding to the 10 highest ranked positive and negative images were estimated, both on average and individually.
Results: :
Subjects judged images blurred by their own HOA as sharper (57% from the presented pairs of images, on average across subjects) than those blurred with 90º rotated version of their HOA (45% of the images). However, in many cases (20 out of 36) the percentage of images considered sharper with other HOA orientations was higher than with the subjects’ own. The classified positive responses correlated more with the subject’s PSF (r=0.46 on average) than the negative (r=0.34) and the difference was significant for all subjects (p<0.02). The positive average PSF showed also a higher correlation with the subject’s PSF than the negative average PSF. The perceptual ranking identified images that were consistently considered blurred or sharp by all subjects.
Conclusions: :
Our results show some bias for blur orientation (particularly their own blur as opposed to 90º rotation), although subjects reveal relatively high tolerance to blur produced by different orientations in the HOA. The codification of internal blur seems to be driven by the overall amount of blur and to a lesser extent, by blur orientation.
Keywords: adaptation: blur • aberrations • perception