Computer Engineering and Applications ›› 2010, Vol. 46 ›› Issue (15): 182-184.DOI: 10.3778/j.issn.1002-8331.2010.15.054

• 图形、图像、模式识别 • Previous Articles     Next Articles

Moving object detection based on Markov Random Field model

JIANG Yong-xin1,JIN Yu-xin2,WANG Xiao-tong3,HUANG Hua3   

  1. 1.Department of Basic Sciences,Dalian Naval Academy,Dalian,Liaoning 116018,China
    2.Department of Computer Science and Technology,Tsinghua University,Beijing 100084,China
    3.Department of Navigation,Dalian Naval Academy,Dalian,Liaoning 116018,China
  • Received:2008-10-13 Revised:2008-12-29 Online:2010-05-21 Published:2010-05-21
  • Contact: JIANG Yong-xin

基于马尔可夫随机场的运动目标检测

蒋永馨1,金俣欣2,王孝通3,黄 华3   

  1. 1.海军大连舰艇学院 基础部,辽宁 大连 116018
    2.清华大学 计算机科学与技术系,北京 100084
    3.海军大连舰艇学院 航海系,辽宁 大连 116018
  • 通讯作者: 蒋永馨

Abstract: Accurate detection of moving objects is an important precursor to stable tracking or recognition.This paper presents an object detection scheme using Markov Random Field and pixel-wise Multi Gaussian Background Model in stationary camera situation.The relation of each pixel and the other neighbor pixel is constituted by Markov Random Field.And the whole restriction is established.This fetches up the shortage of single pixel information with Multi Gaussian Background Model.So the foreground segmentation is more exact.This paper also presents a novel energy function based on Multi Gaussian Background Model and establishes solution scheme using simulated annealing method.Experiments demonstrate its effectiveness.

摘要: 精确的目标检测是目标跟踪和识别的重要前提。提出了一种基于固定摄像机环境下的运动目标检测方案,利用多高斯和马尔可夫随机场的混合模型对视频序列进行前景分割,以达到对运动目标检测的目的。建立了马尔可夫随机场用以刻画图像中每个像素点与一定范围的领域内其他各点的关系,同时考虑一定的时域中的关系从而构建一个全局的约束,弥补多高斯模型只考虑单点信息的不足,使得前景分割更为准确。还给出了一种基于多高斯和马尔可夫随机场的新的能量函数形式,并给出了模拟退火方法对模型进行求解的方法。结果表明,利用该文的方法对运动目标进行检测,结果要优于多高斯模型。

CLC Number: