摘要
背景提取是运动目标检测中重要而基础的一个环节。分析了一般静态背景提取算法的原理和缺陷,提出了一种新颖的基于改进K-均值聚类算法的背景提取方法,给出了动态三元组(DTDG)的概念,并且对每个像素用3个动态三元组进行建模,实现了原始背景的提取。实验验证了所提方案的有效性。
Background generation is an important and basic step in moving objects detection. The analysis of the principle and imper- fection of general methods for stationary background generation is tried. A novel approach to background generation based on modified K-means is proposed. A concept-dynamic-three-dimension-group （DTDG） and pixel modelled using three dynamic-three-dimension- groups are issued out. The method gets the generation of background. The experiment demonstrate the effectiveness of the method.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第21期5141-5143,共3页
Computer Engineering and Design
关键词
背景模型
背景提取
K-均值
均值参考点
计算机视觉
background model
background generation
K-means clustering
mean reference point
computer vision
