作成者 |
|
本文言語 |
|
出版者 |
|
|
発行日 |
|
収録物名 |
|
巻 |
|
開始ページ |
|
終了ページ |
|
出版タイプ |
|
アクセス権 |
|
JaLC DOI |
|
関連DOI |
|
|
|
関連URI |
|
|
|
関連情報 |
|
|
|
概要 |
In the correlation model for stereo vision, local binocular disparity is determined by calculating the value of covariance for small regions called windows between the images of the two eyes. When app...lying the Bayesian approach to stereo vision, it is important to calculate the likelihood of binocular correspondence-the degree to which both eyes are likely to see a single visual scene given a pair of stereo images. This study presents a theoretical ideal to compute the likelihood of binocular correspondence in a correlation model. For several types of stereograms, such as random-dot, grating, and more complex patterns, the likelihood of correspondence was calculate d as a function of the correlation-window size and the range of disparity computation. The results showed that the present model's outputs are consistent with human percepts, suggesting that the idea proposed here is useful in modeling human stereo vision.続きを見る
|