Purchase this article with an account.
Richard Legras, Yohann Benard; Measurement And Prediction Of Through-focus Image Quality Assessment And Depth-of-field In Presence Of Various Combinations Of 4th And 6th Order Zernike Spherical Aberration. Invest. Ophthalmol. Vis. Sci. 2011;52(14):2836.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To explore the through-focus subjective vision (TFSV) (i.e. image quality assessment using a continuous 5-items grading scale) in presence of various single or combinations of monochromatic aberrations using simulated images (i.e. convolution between an original image and calculated PSF) and to compare it with through-focus image quality metrics (TFIQM). To compare the subjective depth-of-field (DoF) based on the bothersome and objectionable blur, with DoF measured on TFSV and on TFIQM.
Five subjects viewed, while dynamically correcting their aberrations, in a first time, diffraction limited image (three 0.4 logMAR letters) and, just after, one of the tested images. The tested images were through-focus (i.e. 0.18 D step) of various aberration conditions: no aberrations with a pupil size of 3 and 6mm, -0.2, -0.4 and -0.6µm of spherical aberration (SA4), +0.2µm of secondary spherical aberration (SA6), and combinations of -0.4µm of SA4 with ±0.2µm of SA6. Subjects had to put a mark for each viewed image on 2 grading scales. The aberrated image score was defined as the difference between the two gradations. All the images were judged 3 times in a random sequence. The subjective DoFs were measured 3 times for each condition. The predicted DoFs were the width of the TFIQM curves measured at various threshold levels.
Adding aberrations enlarged TF curves at the expense of losses in image quality. The largest effect was obtained when combining SA4 and SA6 of opposite signs. The shape of the TFSV were close (r²=0.93, p<0.001) to some of the calculated TFIQMs. The predicted DoFs were respectively measured at 1.05 (i.e. normalization of the IQMs on the 5-items grading scale) and 0.7 to optimize the correlation with the bothersome DoF (i.e. mean absolute difference (AD) of 0.17 D, SD = 0.14 D) and with the objectionable DoF (i.e. mean AD of 0.18 D, SD = 0.14 D). The subjective DoFs respectively measured on the TFSV at 1.15 and 0.3 maximized the correlation with the bothersome DoF (i.e. mean absolute difference (AD) of 0.2 D, SD = 0.12 D) and the objectionable DoF (i.e. mean absolute difference (AD) of 0.18 D, SD = 0.12 D).
TFSV, bothersome and objectionable DoF could be well predicted by some IQMs.
This PDF is available to Subscribers Only