A Task Dependent Model for Visual Performance in Low Light Levels

Abstract

Purpose:: Operating in low light conditions can involve photopic, scotopic, or mesopic vision depending on the light levels and the task at hand. The aim of this research was to derive an applied model of visual performance under low light, primarily mesopic conditions.

Methods:: Eight subjects participated in three low light visual tasks. The first task was central visual acuity on a Bailey Lovie chart. Subjects sat two meters from the chart that was illuminated to mesopic levels with a Hoffman Night Sky projector. The second task measured Landolt C visual acuity on a computer monitor whose luminance was set to 1.7 cd/m². The third task was on- and off-axis visual detection of a 1 cyc/deg Gabor patch on a computer monitor with a luminance of 0.03 cd/m². The target was viewed centrally and peripherally, 30 degrees in the right visual field. For each visual task, subjects wore six neutral density (ND) filters ranging from 10% to 92% photopic luminance transmittance. Scotopic transmission equaled photopic. They also wore seven sets of spectrally selective (SS) filters. In each case, the photopic and scotopic transmittance of each filter was calculated by convolving filter transmittance, the spectrum of the light source, and the CIE photopic and scotopic sensitivity functions.

Results:: The results of the acuity and threshold detection tasks were fit against the transmission of the ND filters. Then the vision task results of the SS filters were fit to the photopic and scotopic transmission of the filters. The R² value of the SS filter data to the fit predicted by the ND filter data was calculated. The optimal mesopic model (MM) predicted by the data was found by simply weighting the photopic and scotopic sensitivity functions with the following equation: MM=αV^/( )+(1-α)V() where α is the relative weight. When α=0 the equation yields the photopic sensitivity function and when α=1 the scotopic function. For Bailey Lovie acuity and on-axis detection, MM provided a better fit than either the scotopic or photopic models. R² values for the MM fits ranged from 0.91 to 0.97. Landolt C acuity was modeled best by the photopic function and off axis detection was modeled best by the scotopic function.

Conclusions:: The applied mesopic model presented here can be used to predict performance for many activities of daily life that are carried out in low light conditions. The weighting function (α) will vary with the specific task parameters.

Keywords: visual acuity • detection