Computer monitor pixellation and Landolt C visual acuity

To investigate the effects of computer monitor pixel density on Landolt C visual acuity measurements and to provide appropriate pixel density recommendations for Landolt C vision chart design.

Participants were 10 healthy observers aged 18-31 years (mean 21.7 ± 3.6). Logarithmic progression charts were used, consisting of eight rows of five Landolt C optotypes ranging from -0.40 logMAR to 0.30 logMAR. Monitor pixel angular subtense varied from 0.10 min arc to 1.97 min arc, achieved by changing the chart test distance. Testing took place with two filtering conditions: unfiltered optotypes, (pixels rendered either black or white) and anti-aliased optotypes with pixel luminance averaged for a pixel square and rendered as grey levels.

Each participant's acuity versus pixel size data sets were fitted by a nonlinear relationship with acuity equal to an asymptotic threshold visual acuity (VA_as ) for small pixel sizes below a critical pixel size (P_crit ). For pixel sizes larger than P_crit  there was a linear relationship between acuity thresholds and pixel size. For anti-aliased Landolt Cs, mean P_crit  was 1.23 min, and for unfiltered Landolt Cs average P_crit was 0.65 min. For anti-aliased LandoltCs, P_crit  was 2.01xVA_as , and for unfiltered Landolt Cs P_crit  was 1.05xVA_as .

These results are quantitatively very similar to previous research on pixellation and Sloan optotype acuity. We have demonstrated that spatially filtering Landolt C optotypes acts as anti-aliasing, to make them more robust to the degradation effects of pixellation. Previous recommendations for maximum pixel size on Sloan letter vision charts can be applied safely to Landolt C charts.