摘要
针对背景的缓变特性和实时图像处理的需求,该文提出了一种基于序列图像时间稳定性特征的背景估计技术。采用一个时间矩阵来监视序列图像的稳定性特征,利用背景在时间上的缓变特性来实现对背景的提取和更新。结果表明:该方法具有简单有效、实时性能好、内存消耗少、背景提取速度快等特点,并对突变背景具有自动感知能力;时间矩阵对于运动目标的分布和轨迹分析也具有参考价值;该方法特别适合静止摄像机的背景估计,在背景配准后也适用于移动摄像机的背景估计。
A background estimation technique was developed based on sequential image time stability for slowly changing background in real-time image processing. A time matrix is used to monitor the sequential image stability, with the background extracted and updated based on the slow changes in the background. The results show that the method can identify an abruptly changing background and still gives fast estimates of the background with good real-time performance, simple algorithms, and low memory cost. The result also shows that the time matrix provides a good reference for analyzing moving objects. The method is very suitable for a stationary camera and is also suitable for a moving camera after background alignment.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2005年第5期601-605,共5页
Journal of Tsinghua University(Science and Technology)
关键词
图像处理
背景估计
序列图像
运动目标
image processing
background estimation
sequential image
moving object
