视频前景多特征融合的人群密度分析

Crowd Density Analysis with Combining Multiple Features of Video Foreground

视频GIS技术可以融合视频处理技术与GIS分析技术实现视频场景的公共安全监控,在公共场所视频监控应用领域,人群密度和数量能够反映地理空间场景的人群拥挤程度,具有一定的公共安全风险控制作用。为满足公共安全领域利用视频GIS技术对视频人群监测及分析的需求,提出了一种视频前景多特征融合的人群密度及数量估计方法。该方法融合前景图像块的边缘纹理特征,通过SVM区分场景人群密度,然后以归一化前景面积、轮廓和边缘纹理特征作为回归因子,根据人群密度差异分类构建回归模型,进而估算场景人群数量。实验结果表明,此方法能够高效估计视频场景的人群密度和数量,准确表征该场景人群分布状态。

Video GIS technology can integrate video processing technology and GIS analysis technology to realize the public security monitoring of video scene. The crowd density and the quantity can reflect the crowds’ crowding degree of the geospatial scene and have certain control effects on public security risk in the field of video surveillance applications in public places. In order to meet the needs of video surveillance and analysis in the security field, this paper proposes a method of crowd density and quantity estimation based on multi-feature fusion of video foreground. The method combines the edge features and texture features of the foreground image block and distinguishes the crowd density in the scene by SVM, then it uses the normalized foreground area, the normalized foreground contour and the edge texture features as regression factorsto construct the regression model according to the crowd density difference classification results and estimates the number of scene crowds. The experimental results show that the method can efficiently estimate the crowd density and quantity in the video scene, and it could accurately represent the distribution status of crowd in the scene.