摘要
针对传统停车场管理方式的不足,提出一种基于以DSP为硬件核心,以颜色矩和区域特征融合的特征匹配算法为软件实现基础的智能网络摄像机实现异常事件检测,包括车辆出入、超速、逆向、违规等,来解决停车场的智能实时监控和管理。实验结果表明,这一改进型算法对异常事件检测有很高的准确性,同时能基本满足系统实时性需求。
According to the deficiency of the traditional parking management mode, a kind of intelligent network camera is put forward which based on DSP core for the hardware and based color moment and regional characteristics of the fusion feature matching algorithm for software. It solves the intelli- gent parking lot real- time monitoring and management by realizing basic exceptional event detection, including car coming in or going out, speeding, driving the reverse direction, parking violation, etc. The experimental results show that the modified algorithm to the abnormal events has high accuracy testing, and can satisfy the system demand of real time.
出处
《电视技术》
北大核心
2012年第11期125-128,147,共5页
Video Engineering
基金
上海大学创新基金项目(A10-0107-09-902)
上海市大学生创新活动计划(CXSJ-11-028)
关键词
DSP
特征匹配
智能监控
停车场
DSP
feature matching
intelligent monitor
parking lot
