The effect of pixel nonlinearity on cathode-ray tube-based visual acuity tests.

Abstract

Pixel nonlinearity of a cathode-ray tube (CRT) display can cause differences between the actual image and the nominal image. One of the discrepancies, the anisotropy of interactions between neighboring pixels on the same raster line and between neighboring raster lines, may have an impact on the CRT-based visual acuity test where small horizontal and vertical gaps are used. We evaluated this impact. Two high-quality color CRT monitors were tested. The stimulus target was a black square ring on a white background. White gaps of one stroke width were opened in the middle of the straight sides of the square ring. In a gap width comparison study, the up and down gaps were flanked with a one-pixel band on the left and right side. The pixel value of the bands was varied to determine the luminance that made up and down gaps appear to have the same width as left and right gaps. The comparison was made at 1 and 2 m. In a visual acuity study, horizontal gap (left/right) acuity and vertical gap (up/down) acuity were measured separately at several distances between 8 and 11 m. Similar measurements were made when the monitor was rotated 90 degrees. Gaps with edges that were parallel to raster lines (raster-parallel gaps) appeared wider than gaps with edges that were perpendicular to raster lines (raster-perpendicular gaps). To match the apparent width of the left and right gaps, the two one-pixel bands flanking the up and down gaps required a pixel value that corresponded approximately to a luminance of mid gray, indicating that the apparent width of a four-pixel left or right gap was similar to that of a five-pixel up or down gap. Raster-parallel gaps produced a higher percentage of correct responses than raster-perpendicular gaps in a visual acuity test. When nominal gap widths were equal, left and right gaps produced 10% or more correct responses than up and down gaps at some target sizes, which translated to a two-letter acuity difference. This acuity difference could not be accounted for by uncorrected astigmatism because the anisotropy persisted when the monitor was turned 90 degrees. Pixel nonlinearity of a CRT-display results in an anisotropy of gap width that can be observed at and above visual acuity size. This anisotropy may introduce uncertainties in the results of CRT-based visual acuity tests where gaps of different orientations are used.