摘要
视频运动对象检测是智能安全监控的基本问题。考虑视频的时间空间分布规律,该文构造了视频六维分布模型以进行运动对象分割。在时间维上利用在线高斯混合模型对背景进行更新,并且计算像素颜色差异。在空间维上利用纹理特征描述像素点之间结构关系,提高了对光线变化和阴影等环境变化的抗干扰能力。实验结果表明该算法是实用有效的。
Video motion object detection is a foundation of aptitude security surveillance. Considering the spatial-temporal distribution of video, a six dimension video distribution （SDVD） is constructed for segmentation of motion object. The on-line Gaussian mixture model is employed to update the background image and to calculate for every specified pixel color difference in temporal dimension. Texture feature is also used to describe the local structure information of pixel neighbor in spatial dimension, so as to enhance the robustness of light change and shadow existence. The experiment results show that the proposed SDVD is effective and practical.
出处
《计算机工程》
CAS
CSCD
北大核心
2007年第10期227-229,共3页
Computer Engineering
基金
广东省重点攻关基金资助项目(2004B10101032)
关键词
视频分割
视频六维分布模型
纹理特征
在线高斯混合模型
Video segmentation
Six dimension video distribution （SDVD）
Texture feature
On-line Gaussian mixture model
