摘要
阐述了利用面向视频智能分析技术,对商业区行人交通流数据进行样本提取、预处理和建模分析的方法全过程。以北京西单商业区为例,构建了包含不同类型监测点、不同时间点的日期分组式纵向时间序列,并完成了预测建模和效果对比。研究表明,所有序列均为平稳非白噪声序列,具有相似的自回归移动平均(ARMA)模型形式,能较好地实现对行人流量的预测。
The sampling,the preprocessing,and the modeling for the commercial street pedestrian traffic flow data based on the intelligent video analysis are presented in this paper.The Xidan mall is taken as an example,the by-date grouping-style vertical time series are established,consisting of multi surveillance points and different points-in-time.The modeling and forecast results show that all vertical time series are stationary and of non-white noise with similar ARMA expression formulas,which can be well applied to the forecast of pedestrian traffic flow data.
作者
倪慧荟
吴波鸿
NI Huihui;WU Bohong(Research Laboratory of Safety and Emergency Management,Beijing Municipal Institute of Labour Protection,Beijing 100054,China;Institutes of Science and Development,Chinese Academy of Sciences,Beijing 100190,China;University of Chinese Academy of Sciences,Beijing 100049,China)
出处
《科技导报》
CAS
CSCD
北大核心
2019年第16期74-82,共9页
Science & Technology Review
基金
北京市科学技术研究院创新工程预研项目(PXM2018_178304_000011)
关键词
视频智能分析
商业街行人交通流
预测建模
intelligent video analysis
commercial street pedestrian traffic flow
forecast model
