一种基于局域灰度分布的自适应运动检测算法

运动检测是基于视频的智能交通系统的基础,为了能够快速、准确的适应背景变化,本文提出了一种基于局域灰度分布的运动检测算法,利用图像块的灰度分布特征,构造相应的离散概率模型,并采用自适应的模型参数更新方法。实验证明,算法能够自适应交通场景中背景的多模态性、动态背景、光照变化以及一定程度的遮挡等情况,并且训练时间短,每帧的处理时间也能够满足实时性要求。

An Adaptive Algorithm Based on Local Grey-level Distribution for Motion Detection

Motion detection is an essential step in the intelligent traffic system based on video.In order to correctly deal with vari- ances of background quickly,this paper'proposes a novel adaptive motion detection algorithm based on local grey-level distribution.It constructs the discrete probabilistic model for each image block and adaptively updates the parameters of the model.Experiments show that this method can effectively handle the practical difficulties in traffic scenes,including multimodal background,dynamic background, illumination changes,and partial occlusion.In addition,this algorithm requires less training samples and can satisfy real time requirement of processing time per frame.