PubMed Health⌕ Search
Home/Archive/PubMed

PubMed · 8935905

Structure from motion: a tolerance analysis.

M A Hogervorst·A M Kappers·J J Koenderink
View original source ↗Export metadata ↓

Abstract

We present a tolerance analysis that is applicable to a large group of stimuli used in structure-from-motion tasks. Human performance in structure-from-motion tasks reflects the fact that the visual system deals with projections of a 3-D world on the retina. A tolerance analysis reveals the relationship between the projections and the 3-D world. Any realistic model of the visual system should incorporate a tolerance analysis as a complete description of the stimulus. By way of example we apply the tolerance analysis to the stimuli used in two widely known experiments in which different properties of structure were tested--that is, perceived nonrigidity (Norman & Todd, 1993) and ordering in depth (Hildreth, Grzywacz, Adelson, & Inada, 1990). The analysis explains qualitatively the results of these experiments, illustrating that the results are to a large extent due to stimulus limitations rather than to mechanistic properties of the visual system. From our analysis it follows that far more sensitive measurements of the optic information are needed to obtain metric structure than affine structure.

Explore related subjects

Depth PerceptionDiscrimination LearningHumansMotion PerceptionOrientationPattern Recognition, VisualProblem SolvingPsychophysicsJournal ArticleResearch Support, Non-U.S. Gov't

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

M A Hogervorst, A M Kappers, J J Koenderink. 1996. Structure from motion: a tolerance analysis.. https://doi.org/10.3758/bf03206820

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed1999

The popout in some conjunction searches is due to perceptual grouping.

The target in a visual search task usually pops out if it can be distinguished from its background on the basis of only one visual feature but not if the target represents a conjunction of two or more features. However, several recent reports suggest that in certain cases, search targets defined by a conjunction of two features also pop out. We have reinvestigated three pairs of such features to determine whether the popout in these cases can be attributed to perceptual grouping. We find that that in all three cases, popout no longer occurs when perceptual grouping is degraded, suggesting that the popout is the result of perceptual grouping and not of novel mechanism/s of conjunction search.

Depth Perception↗
PubMed1999

Variational learning in nonlinear gaussian belief networks.

We view perceptual tasks such as vision and speech recognition as inference problems where the goal is to estimate the posterior distribution over latent variables (e.g., depth in stereo vision) given the sensory input. The recent flurry of research in independent component analysis exemplifies the importance of inferring the continuous-valued latent variables of input data. The latent variables found by this method are linearly related to the input, but perception requires nonlinear inferences such as classification and depth estimation. In this article, we present a unifying framework for stochastic neural networks with nonlinear latent variables. Nonlinear units are obtained by passing the outputs of linear gaussian units through various nonlinearities. We present a general variational method that maximizes a lower bound on the likelihood of a training set and give results on two visual feature extraction problems. We also show how the variational method can be used for pattern classification and compare the performance of these nonlinear networks with other methods on the problem of handwritten digit recognition.

Depth Perception↗
PubMed1999

Occlusion contributes to temporal processing differences between crossed and uncrossed stereopsis in random-dot displays.

Stereoscopic depth discrimination was investigated in crossed and uncrossed directions using stimuli defined by binocular disparity differences embedded in dynamic random-dot stereograms. Across three experiments, fixation was directed to a point on the display screen (which placed crossed stimuli in front of and uncrossed stimuli behind, the background dots of the stereogram), to a point in front of the display screen (which placed both crossed and uncrossed stimuli in front of the background dots), and to a point behind the display screen (which placed both crossed and uncrossed stimuli behind the background dots). Results showed that depth discrimination was always good when the stimuli appeared in front of the background dots of the stereogram, whereas discrimination was always poor when the stimuli appeared behind the background dots. These results suggest that differences between crossed and uncrossed stereopsis as reported in past research arose, in part, from effects related to occlusion.

Depth Perception↗