孔素然. 基于归一化前景和二维联合熵的人员非法聚集图像检测[J]. 微电子学与计算机, 2015, 32(8): 139-141,145. DOI: 10.19304/j.cnki.issn1000-7180.2015.08.029
引用本文: 孔素然. 基于归一化前景和二维联合熵的人员非法聚集图像检测[J]. 微电子学与计算机, 2015, 32(8): 139-141,145. DOI: 10.19304/j.cnki.issn1000-7180.2015.08.029
KONG Su-ran. The Personnel Illegal Gathering Image Detection Based on the Normalized Prospects and Two-dimensional Entropy[J]. Microelectronics & Computer, 2015, 32(8): 139-141,145. DOI: 10.19304/j.cnki.issn1000-7180.2015.08.029
Citation: KONG Su-ran. The Personnel Illegal Gathering Image Detection Based on the Normalized Prospects and Two-dimensional Entropy[J]. Microelectronics & Computer, 2015, 32(8): 139-141,145. DOI: 10.19304/j.cnki.issn1000-7180.2015.08.029

基于归一化前景和二维联合熵的人员非法聚集图像检测

The Personnel Illegal Gathering Image Detection Based on the Normalized Prospects and Two-dimensional Entropy

  • 摘要: 提出一种基于归一化前景和二维联合熵的人员非法聚集图像检测方法,将若干高斯模型依据可靠性度量按照降序排列,用排在前面的高斯分布描述背景分布.将前景面积当成人员非法聚集图像检测模型中的一个参数.对原有前景面积公式进行归一化处理,通过二维联合概率密度运算二维联合熵.将场景划分成网格状统计区间,计算所有统计区间的前景概率密度,结合香农信息熵理论,求出前景的二维联合分布熵,得到人员非法聚集检测模型.引入遮挡因子对其进行调整,调整体现场景人员非法聚集程度的参考指数并设定其阀值,确定场景是否发生大规模人员非法聚集.仿真实验结果表明,所提出的方法具有较高的检测精度.

     

    Abstract: A kind of personel illegal gathering imagine detection is given out in this paper. Some Gaussian models were obtained in a descending order according to the reliability measurements. The front Gaussian distribution was used to describe the background distribution. The prospect area was thought as a parameter of the personel illegal gathering image detection model. The original prospect area fornula was normailzed, and 2 D entropy was caculated by the 2D joint probability density. The scene was divided into some grid statistical intervals. All prospect probability density was caculated. On the basis of Shannon entropy theory, the 2D joint distribution entropy was obtained and the peronel illegal gathering detection model was constructed. The simulation results show that the proposed method is of high precision.

     

/

返回文章
返回