基于背景差法的几种背景建模方法的研究

目前最常用也最有效的运动目标检测方法是背景差法,其中背景提取是背景差法的核心。主要比较了均值滤波法、中值滤波法、混合高斯分布法三种常用的背景建模的方法,说明了各算法在不同情况下的性能优劣。通过实验发现,均值滤波法和中值滤波法适合特定的场合,混合高斯法在场景复杂的情况下效果更好。

Research of Several Background Modeling Based on Background Subtraction

The most common and effective approach in moving object detection is background subtraction,and the core of the is background modeling.This paper compare several common methods of background modeling and their advantages and disadvantages in different conditions.The results show that mean filter and median filter are applicable in given conditions,the mixture of Gaussian is better in complex situations.