摘要
针对混合高斯模型的初始建模速度慢,检测出的运动目标含大量阴影和频繁闪动等问题,提出了一种融合背景减除法的改进混合高斯算法。该算法在初始建模时,自适应地更新均值和方差,能快速准确地建立背景模型;结合背景减除法,克服频繁闪动,抑制阴影。实验结果表明,该算法在初始建模、运动目标检测效果等方面优于混合高斯算法,具有较强的稳定性和适应性。
Aiming at the shortcoming of mixture Gaussian model which has a lowly initial modeling speed, and in which the detected moving targets have lots of shadows and the frequent flashing, it proposes an improved mixture Gaussian algorithm with the fusing background subtraction. This algorithm adaptively updates the mean and variance in the initial modeling, which can build the background model quickly and accurately. It is combined with background subtraction algorithm to overcome the frequent flashing, and to suppress shadows of moving target. Experiments show the proposed algorithm, with strong stability and adaptability, can be superior to the mixture Gaussian model at aspects of the initial modeling and detecting results.
出处
《计算机工程与设计》
CSCD
北大核心
2011年第2期592-595,共4页
Computer Engineering and Design
基金
国家自然科学基金项目(10676029
10776028)
四川省教育厅重点基金项目(2006C074
2006C075)
关键词
运动目标检测
混合高斯模型
背景减除法
初始速度
阴影
闪动
moving object detection
mixture of Gaussian model
background subtraction
initial modeling speed
shadow
flashing