Perceptual grouping as Bayesian estimation of mixture models V Froyen, J Feldman, M Singh |
|---|
Dept. of Psychology, RuCCS, Rutgers University - New Brunswick, NJ, United States
|
| We propose a Bayesian approach to perceptual grouping in which the goal of the computation is to estimate the organization that best explains an observed configuration of image elements. We formalize the problem as a mixture estimation problem, where it is assumed that the configuration of elements is generated by a set of distinct components ("objects"), whose underlying parameters we seek to estimate (including location and "ownership" of image elements). An important aspect of this approach is that we can estimate the number of components in the image, given a set of assumptions about the underlying generative model. We illustrate our approach, and compare it to human perception, in the context of one such generative class: Gaussian dot-clusters. In two experiments, we showed subjects dots that were sampled from either two (Exp. 1) or three Gaussian clusters (Exp. 2). In both experiments we manipulated the distances between the clusters in order to modulate the apparent number of clusters. Subjects were asked to indicate how many clusters they perceived. We found that numerical estimates based on our Bayesian model closely matched subjects' responses. Thus our Bayesian approach to perceptual grouping, among other things, effectively models the perception of cluster numerosity. |
| Up Home |
|---|