摘要
交通流量的视频检测离不开背景提取。现有的背景提取算法往往运算量大,不能快速跟踪背景变化。本文提出一种基于区间分布的自适应背景提取算法,该算法利用背景像素值的分布特征,得到背景的估计模型。实验结果表明,相对于其他自适应背景提取算法,区间分布法能够较好地自适应环境光线的变化,快速跟踪背景的变化,同时计算量较小,能够满足实时性要求。
Background abstraction is an essential step of traffic flow detection based on video. Currently, background abstraction algorithm with high computation cost can not deal with fast variance of background. This paper proposes a new adaptive background abstraction algorithm based on Zone--Distribution. The algorithm constructs the background estimation model based on the distribution characteristics of background pixel value. The experimental results show that the al-- gorithm can be adaptive of variance of environment lighting condition and track background change. The computation cost of the Zone--Distribution algorithm over other adaptive algorithms is smaller and can satisfy the real--time requirement.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2005年第3期316-321,共6页
Pattern Recognition and Artificial Intelligence
关键词
智能交通系统
交通流量检测
背景提取
Intelligent Transportation System
Traffic Flow Detection
Background Abstraction
