Motion Integration Using Competitive Priors
Shuang Wu,Hongjing Lu,Alan L. F. Lee,Alan L. Yuille +3 more
- 24 Jul 2009
- pp 235-258
TL;DR: This work defines novel prior models for smooth rotation and expansion using techniques similar to those in the slow-and-smooth model [23] (e.g. Green functions of differential operators).
read more
Abstract: Psychophysical experiments show that humans are better at perceiving rotation and expansion than translation [5][9]. These findings are inconsistent with standard models of motion integration which predict best performance for translation. To explain this discrepancy, our theory formulates motion perception at two levels of inference: we first perform model selection between the competing models (e.g. translation, rotation, and expansion) and then estimate the velocity using the selected model. We define novel prior models for smooth rotation and expansion using techniques similar to those in the slow-and-smooth model [23] (e.g. Green functions of differential operators). The theory gives good agreement with the trends observed in four human experiments.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Figures
Fig. 5. Stimulus and results in Experiment 2 with randomly-oriented grating stimuli. Left panel: illustration of grating stimulus. Blue arrows indicate the drifting velocity of each grating. Middle panel: human coherence thresholds for different motion stimuli. Right panel: Model prediction of coherence thresholds which are consistent with human trends. 
Table 1. Model selection result for the grating rotation stimuli. The values are logarithms of model evidence. The correct model, rotation model, always wins. As the coherence ratio increases, the rotation model’s advantage also increases. 
Table 2. Model selection result for the grating expansion stimuli. The correct model, expansion, always wins. 
Table 3. Model selection result for the grating translation stimuli. All three models (rotation/expansion/translation) have virtually the same model evidence. This is due to the fact that rotation/expansion models also favor translation as translation model does. 
Fig. 1. An illustration of observing a walker with a moving camera. Top panel, three example frames. Bottom panel, observing the scene through a set of punch holes 
Fig. 7. Results in Experiment 4 with grating stimuli to compare rigid versus nonrigid rotation and expansion. Left panel: human coherence thresholds for rigid and non-rigid conditions as a function of different motion patterns. Right panel: Model prediction of coherence thresholds which are consistent with human trends.
Citations
Perceptual learning improves discrimination while distorting appearance
TL;DR: It is found that while training improves discrimination ability, it leads to increases in appearance distortion, and a model of how distortions of appearance can arise from increased precision of neural representations and serve to enhance distinctions between perceptual categories is proposed.
3
Causal Context Presented in Subsequent Event Modifies the Perceived Timing of Cause and Effect
TL;DR: Novel causal displays in which various types of causal contexts could be presented in subsequent events were used, showing that the perceived causal context modulated the temporal relationship of two serial events so as to be consistent with the causal order implied by the subsequent event.
3
References
Determining optical flow
TL;DR: In this paper, a method for finding the optical flow pattern is presented which assumes that the apparent velocity of the brightness pattern varies smoothly almost everywhere in the image, and an iterative implementation is shown which successfully computes the Optical Flow for a number of synthetic image sequences.
12.4K
•Book
Determining optical flow
Berthold K. P. Horn,Brian G. Schunck +1 more
- 03 Jan 1992
TL;DR: An iterative implementation is shown which successfully computes the optical flow for a number of synthetic image sequences and is robust in that it can handle image sequences that are quantified rather coarsely in space and time.
The Robust Estimation of Multiple Motions
TL;DR: A framework based on robust estimation is presented that addresses violations of the brightness constancy and spatial smoothness assumptions caused by multiple motions of optical flow, and is applied to standard formulations of the optical flow problem thus reducing their sensitivity to violations of their underlying assumptions.
1.9K
Stevens' handbook of experimental psychology
Hal Pashler
- 15 Jul 2002
TL;DR: Vol. 1 1. Neural Basis of Vision, 2. Associative Structures in Pavlovian and Instrumental Conditioning, 3. Reinforcement Learning, and 4. Learning: Laws and Models of Basic Conditioning.
1.6K
•Book
Perception as Bayesian Inference
David C. Knill,Whitman Richards +1 more
- 01 Sep 1996
TL;DR: In this article, a Bayesian decision theory for binocular stereopsis is proposed based on the notion of shape from texture, and the generic viewpoint assumption in a Bayes framework.
1.4K