视频图像理解在客流统计中的应用  

Application of video image understanding to measurement of pedestrian flow

作　　者：梁英宏[1] 王知衍[1] 曹晓叶[1] 许晓伟[1] 

机构地区：[1]华南理工大学计算机科学与工程学院,广东广州510640

出　　处：《计算机工程与设计》2008年第5期1203-1206,共4页Computer Engineering and Design

基　　金：科技部科技型中小型企业技术创新基金无偿项目(02C26214400224);广东省科技计划基金项目(2002A1020104)

摘　　要：视频图像理解侧重于对时间序列进行分析,既涉及到图像的空间特性,也涉及到视频序列的时间特性,是目前计算机视觉领域的一个研究热点,而客流统计是视频图像理解的一个重要应用。提出了视频图像理解的层次结构,即视频图像分割、目标识别、场景和行为理解,同时对每个层次的操作对象、任务和技术领域进行描述,然后总结了客流统计方法的一些研究成果,将客流统计分成行人计数和人群密度估计两种问题,并利用该层次结构解决这个实际应用。Video image understanding technology, as a hotspot in computer vision, places emphasis on interpreting video sequences, which involves the space property of images and the timing feature of video sequences. Pedestrian flow measurement is an important application in the field of video image understanding. First, a layered framework for video image understanding is presented, including video image segmentation, object recognition and scene/behavior understanding. Meanwhile, the operational objectives, tasks and related technical fields of each layer are discussed. Then, the current research status of pedestrian flow measurement is summarized. Finally, the problem of pedestrian flow measurement is divided into two and solved by using the proposed layered framework, problems： Pedestrian counting and crowd density estimation.

关 键 词：客流统计 视频图像理解 视频图像分割 行人计数 人群密度估计 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]