Determining mesopic visual salience thresholds in see-through displays

Purpose : Unlike in bright (photopic) environments the limits of how much larger or brighter a target amongst distractors needs to be to reliably attract attention in dim (mesopic) environments is not well known. Understanding the limits of human mesopic vision has implications for see-through aviation displays, such as Helmet Mounted Displays (HMDs), as information demands increase. This study conducted a series of repeated measures visual psychophysical experiments to establish visual salience thresholds for size and luminance of targets in mesopic conditions compared to photopic conditions. It was hypothesised that size and luminance thresholds in mesopic conditions will be higher than photopic light conditions.

Methods : Participants (20-33 years, 6/6 vision in both eyes unaided) undertook two computer-based visual search experiments which involved detecting a target among an array of distractors differing by their position and set size (8,16). Distractors were 4.3 arcminutes in size, similar to current see-through display symbology. The target was presented on 50% of trials and differed from the distractors by either stroke width (Expt 1) or luminance against a low contrast background (Exp 2). Participants indicated with a button press if the target was present. Three light levels were tested: Photopic (47 cd/m2), High Mesopic (4.7 cd/m2), Mid Mesopic (0.5 cd/m2). Visual thresholds for each feature for each light level were determined by an 1up-4-down staircase procedure. Threshold estimates across light levels were analysed by 2-way mixed ANOVAs.

Results : For a target to be detected (Exp 1) the increase in stroke width compared to that of distractors must be 1.25 arcmin for photopic, 2 arcmin for high mesopic and 2.75 arcmin larger (N = 5, p < .05) for mid-mesopic light level. In Exp 2, against a low contrast background the absolute luminance of the target needed to be 23% brighter for photopic, 30% for high mesopic and 42% for mid mesopic (N =5, p < .05).

Conclusions : In mesopic light levels both stroke width and luminance of target stimuli must be increased to match detection accuracy in photopic light levels. The results of this research provide empirical limits for human vision in dim light conditions which can be utilised in future visual displays in aviation environments to improve performance.