Perceptual stimulus――A Bayesian-based integration of multi-visual-cue approach and its application |
| |
基金项目: | Supported by the National Natural Science Foundation of China (Grant Nos. 60405004 & 60635050) and National Hi-Tech Project (Grant Nos. 2006AA01Z318 & 2006AA01Z192) |
| |
摘 要: | With the view that visual cue could be taken as a kind of stimulus, the study of the mechanism in the visual perception process by using visual cues in their probabilistic representation eventually leads to a class of statistical Integration of multiple visual cues (IMVC) methods which have been applied widely in perceptual grouping, video analysis, and other basic problems in computer vision. In this paper, a survey on the basic ideas and recent advances of IMVC methods is presented, and much focus is on the models and algorithms of IMVC for video analysis within the framework of Bayesian estimation. Furthermore, two typical problems in video analysis, robust visual tracking and “switching problem” in muIti-targèt tracking (MTT) are taken as test beds to verify a series of Bayesian-based IMVC methods proposed by the authors. Furthermore, the relations between the statistical IMVC and the visual perception process, as well as potential future research work for IMVC, are discussed.
|
关 键 词: | 感知刺激 贝氏统计 视像分析 视觉冲击 |
收稿时间: | 2007-08-31 |
修稿时间: | 2007-12-25 |
Perceptual stimulus — A Bayesian-based integration of multi-visual-cue approach and its application |
| |
Authors: | JianRu Xue NanNing Zheng XiaoPin Zhong LinJiang Ping |
| |
Institution: | (1) Institute of Artificial Intelligence and Robotics, Xi’an Jiaotong University, Xi’an, 710049, China |
| |
Abstract: | With the view that visual cue could be taken as a kind of stimulus, the study of the mechanism in the visual perception process by using visual cues in their probabilistic representation eventually leads to a class of statistical integration of multiple visual cues (IMVC) methods which have been applied widely in perceptual grouping, video analysis, and other basic problems in computer vision. In this paper, a survey on the basic ideas and recent advances of IMVC methods is presented, and much focus is on the models and algorithms of IMVC for video analysis within the framework of Bayesian estimation. Furthermore, two typical problems in video analysis, robust visual tracking and "switching problem" in multi-target tracking (MTT) are taken as test beds to verify a series of Bayesian-based IMVC methods proposed by the authors. Furthermore, the relations between the statistical IMVC and the visual per- ception process, as well as potential future research work for IMVC, are discussed. |
| |
Keywords: | perceptual stimulus integrating multiple visual cues Bayesian estimation video analysis visual tracking |
本文献已被 维普 SpringerLink 等数据库收录! |
| 点击此处可从《中国科学通报(英文版)》浏览原始摘要信息 |
| 点击此处可从《中国科学通报(英文版)》下载免费的PDF全文 |