Abstract
Purpose: :
To evaluate a new, free-recall-based measure of information acquisition from video, by testing its ability to distinguish the effects of different levels of blur, a simple simulation of low vision.
Methods: :
Twenty video clips were chosen from the set of 200 for which we had previously collected baseline responses from normally-sighted subjects. In two experiments, subjects viewed these clips in random order, and their free-recall descriptions of the clips were scored by counting the number of words shared with the baseline responses. The first experiment tested the effect of video degradation produced by image processing with Gaussian low-pass filters. Three levels of blurring were included: no blurring, medium and severe blurring (corresponding to filter standard deviations of 0%, 2% and 6% of the movie width). The blur filters produced visual acuities of about 20/350 and 20/1100 for typical viewing conditions. Mechanical Turk workers (N = 49) provided written responses. The second experiment tested the effect of vision degradation. Participants in the lab (N = 5) viewed the original clips through lenses that produced spherical defocus blur of +3 to +12D, comparable to the image-processing blur. They provided verbal responses.
Results: :
Greater blur produced a greater reduction in the amount of information that was acquired, a dose response effect. Repeated-measures ANOVAs and follow-up pairwise tests showed that the average free-recall scores were significantly different between each of the blur levels, p < 0.001. Results from the defocus blur and the image-processing blur experiments were comparable.
Conclusions: :
The free-recall measure of information acquisition was correlated with the level of blur applied to the video. Therefore, this objective measure can be applied to assess impacts of low vision on video viewing, and effects of video transmission limitations (e.g. due to video compression artifacts) for normally-sighted viewers.
Keywords: low vision • clinical research methodology • clinical (human) or epidemiologic studies: systems/equipment/techniques