一种自适应的背景模型建立方法

在背景差分法的运动目标监测中,背景通常用前m帧图像的平均值来估计,m的取值决定着背景估计的准确性,该值太大实时性差,太小准确性差.本文提出一种自适应估计最小m值的方法,先计算不同m取值下估计背景的标准差δ(m);然后寻找δ(m)序列的第1个最小值,对应的m值为估计背景所需最少帧数,此时背景为最佳背景模型.实验结果表明该方法对不同环境下获取的图像序列有较好的适应性,能使用最少帧数建立背景模型,为后续的运动目标检测奠定基础.

An Adaptive Background Model Method

When we use the background difference method to monitor the moving target,it is usually with the mean value of pre-m frames to estimate the background,and the estimated value of m determines the accuracy of the background.If the value of m is too large the real-time performance is poor,or else the accuracy is poor.This paper presents an adaptive method to estimate the minimum m.Firstly,calculate the standard deviation δ(m) of the background under the different m values;Then search for the first minimum of δ(m) the sequence,the corresponding value of m is the necessary minimum number of frames for estimating the background,so the background is the best background model.The experimental results show that the method has a better adaptability to the obtained image sequence under different environments,which can use a minimum number of frames to establish the background model and lay the foundation for moving target detection for the follow-up.