Purchase this article with an account.
Athena D. Buckthought, Lew B. Stelmach; Spatial scale interactions in stereopsis for different types of band-limited stimuli. Journal of Vision 2002;2(7):328. doi: 10.1167/2.7.328.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Stereo depth perception depends on computing the correspondence between image features in the two eyes. This computation is believed to occur in parallel at different spatial scales of the stimulus, and to yield independent estimates of disparity at each scale. In simple terms, the input image is filtered into spatial frequency bands and binocular disparities are computed separately for each band. In order to characterize these spatial-frequency selective computations, various studies have utilized band-limited stimuli such as Difference-of-Gaussian (DOG), Gabor and filtered noise patterns. Results have generally confirmed the importance of spatial scale in stereo depth perception, and have also identified interactions among spatial scales, but have yielded some puzzling inconsistencies. In particular, results have shown that factors other than spatial scale, such as the envelope or outline of the stimulus can play a role in reducing false matches in repetitive and high spatial frequency stimuli. Following this line of reasoning, our research compared how the spatial frequency of different types of band-limited stimuli affected performance in a depth discrimination task, where the pedestal disparity of the reference stimulus was varied. Our immediate goal was to explain why DOG stimuli used by Siderov and Harwerth (1993) revealed a minor effect of spatial frequency, whereas filtered noise used by Smallman and MacLeod (1997) revealed a very large effect. We confirmed that the differences found using DOG and filtered noise patterns were reliable. We also conducted an experiment measuring depth discrimination using filtered noise that was contrast modulated with a Gaussian envelope, thus bridging the effects observed with DOG and filtered noise stimuli. We explained the differences between the different stimuli using a model that was sensitive to the number of false matches in the stimulus.
This PDF is available to Subscribers Only