基于均匀混合模型的运动检测

Motion detection with uniform mixture model

文章提出一种新的实时运动检测方法,即均匀混合模型运动检测法。该方法对每个象素用均匀混合分布建模,然后利用自适应的学习率在线更新模型,使背景模型更加准确。均匀混合模型运动检测方法显著的特点是,首先模型简单实用,适于运动检测的实时处理;其次利用帧间信息对学习率进行自适应调整,使算法在复杂情况下也能进行正确的运动检测。实验表明该方法较目前常用的高斯混合模型运动检测法有更好的实时性和可靠性。

In this paper a novel approach for real-time motion detection is presented,which is called motion detection with uniform mixture model.In this approach,each pixel is modeled by a mixture of uniforms.The uniform distributions are then updated with adaptive learning rate.The model is simple and utility,and the learning rate is adaptive with different environments which are the salient characteristics of this approach.Experiments show this method is more robust and accurate than the Gaussian mixture model.which is the common approach for real-time motion detection at present.