.png){kind=icon}
返回
Distributed data fusion for real-time crowding estimation
.png){kind=icon}
PRE
AI
DOI：10.1016/0165-1684(96)00075-8.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
This paper describes an approach to estimating the number of pedestrians present in a real-life complex scene by processing a sequence of images. A set of features extracted from each acquired image is related to the number of people monitored. The static declarative knowledge describing the correlation between a feature value and the estimate of the number of people is represented by nonlinear piecewise models. The proposed estimation approach exploits temporal information by means of a distributed Kalman filter network. The current crowding estimate is derived from the application of the static correlation models to the extracted observations (evidence), combined with the prediction obtained from the temporally previous estimate (expectation). This approach improves the estimation accuracy obtained by using only current available data (i.e., static knowledge), and can predict crowding values between two successive acquisitions, i.e., at those steps at which new data cannot be available. Extensive experimental tests carried out in the field (a metro station) yielded very promising results in terms of estimation accuracy and real-time response capability.
Keyword:
crowding estimation
distributed data fusion
distributed extended Kalman filtering
nonlinear static and dynamic knowledge modelling
AI总结
对已上传原文的论文进行重点信息的提取，主要内容包括：简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。
期刊
IF:
3.6
论文数:
10.0K
被引数:
1.7W
机构
暂无机构信息
引用论文
.png){kind=icon}
查看更多内容