基于视频传感器网络和交通信号的路口前景检测  

Intersection foreground detection based on video sensor networks and traffic signal

作　　者：戈军[1] 周莲英[2] 

机构地区：[1]宿迁学院计算机科学系,江苏宿迁223800 [2]江苏大学计算机科学与通信工程学院,江苏镇江212013

出　　处：《计算机应用研究》2015年第7期2214-2217,2221,共5页Application Research of Computers

基　　金：宿迁学院重点科研基金资助项目(2013KY15)

摘　　要：针对传统背景减法并不完全适合路口前景检测的需要,提出一种利用交通信号增强背景减法性能的前景检测新方法。该方法将交通信号与视频传感器网络相结合,通过传感器网络感知环境变化,从而获取实时准确的交通视频信号,并为各像素分配自适应学习率。新旧方法的对比测试实验结果表明,新方法提高了检测精度,具有广阔的应用前景。In view of the traditional background subtraction is not completely suitable for intersection foreground detection needs, this paper proposed a new foreground detection method to enhance the performance of background subtraction using traffic signal. The method combined the traffic signals and video sensor networks ,through the sensor networks to sense environmental changes so as to obtain accurate real-time traffic video signal, and assigned an adaptive learning rate to each pixel. The new and old method contrast test results show that the new method improves the detection accuracy, and the method has broad application prospect.

关 键 词：高斯混合模型 视频传感器网络 前景检测 背景减法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]