摘要
背景减除法是运动目标检测的常用方法,其性能取决于所使用的背景模型。文中针对混合高斯模型不能应对光线突变的问题,提出了一种改进的背景模型。首先选择了新的模型参数,并对模型的更新机制进行了改进,使用了固定的学习率且对方差的更新加入了自适应的更新因子,使其可以适应局部的快速光照变化;其次对模型加入了帧间处理使其可以适应全局的快速光照变化。实验表明,改进的方法能适应各种条件下的光照变化,提高了运动目标检测的精确度。
Background subtraction is one of the common methods in motion detection,its performance depends on the background model.This paper proposes an improved background subtraction method based on Gaussian mixture background model which can not deal with the problem of scene light rapid change.First,a new model parameters was selected,and model updating mechanism was improved by using a fixed learning rate and adding the adaptive factor to update the varience,it can adapt to local illumination changes rapidly.Then,the frame disposal was joined to the model so that it can adapt to deal with the rapid global changes in light.Experimental results show that the improved method can adapt to changing light conditions,and improve motion detection precision.
出处
《计算机技术与发展》
2011年第2期140-142,146,共4页
Computer Technology and Development
基金
国家自然科学基金面上项目(60778007)
关键词
光照变化
背景减除
混合高斯模型
帧间处理
light change
backgroung subtraction
Guassian mixture model
frame disposal
