基于视频的行人检测和行为识别研究

项目名称： 基于视频的行人检测和行为识别研究

项目编号： No.61271288

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 无线电电子学、电信技术

项目作者： 马争

作者单位： 电子科技大学

项目金额： 88万元

中文摘要： 行为识别是通过一系列的观测证据（图像或视频）识别一人或多人的动作行为和动机。该研究领域因为其广阔的应用前景获得了极高的关注度。如今它主要的应用领域包括辅助照顾病人和残疾人，视频监控，人机交互控制，游戏等等。本项目主要研究基于视觉的行为识别算法包括：行人检测和分割，基于高层语义特征的行为识别算法，复杂周期运动识别的难点问题。另外项目将建立一个完备的行为识别数据库。考虑在多个视角，不同背景环境和不同光照条件下采集人体行为图片。可以预计本项目的研究成果能够较大的推进我国在该领域的发展，也将会有很广阔的应用市场和前景。

中文关键词： 行人检测；目标分割与跟踪；行为识别；复杂周期运动；

英文摘要： Activity recognition aims to recognize the actions and goals of one or more persons from a series of observations on the persons' actions and the environmental conditions.Now this research field has captured a lot of attention due to its bright application prospects, examples include assisting the sick and disabled, monitoring video automatically, interacting between human and computer,etc.The main research of this project is vision-based activity recognition including person detection and segmentation,high-level semantics feature based activity recognition, many central issues of complicated periodic motion.In addtion, we will build a complete activity database which obtain activity images and videos from different angel of view, different circumstances, different light conditions,etc.We can expect that the reearch results of this project can greatly promote the developmant of activity recognition flied in our country.

英文关键词： Pedestrian detection；object segmentation and tracking；behavior recognition；complex periodic motion；