基于融合判决的人群密度分析方法研究
rowd density estimation based on comprehensive judgment
针对视频监控中人群密集程度的估计问题，综合运用基于前景像素数目的分析方法和基于图像纹理模式的分析方法，并提出了一种融合判决准则。在前景分析中，提出了对前景图像进行边缘提取，并经过透视视角归一化处理，将前景边缘像素数目与人群密集度的整体类别做映射的方法。在纹理分析中，提出了对视频图片进行重点区域提取的预处理方法，将纹理特征统计量与对应的不同人群密集度建立映射关系。实验结果表明，基于融合的判决准则，对监控场景的人群密度估计具有良好效果。
In this paper a comprehensive judgement method, considering both the method based on the foreground pictures and that based on texture information, is proposed for crowds estimating in video surveillance. Forground-based method obtains the eduge picture elements of foreground pictures, normalizes the features for perspective, and uses the mapping between edge elements and different classes of crowd density. In the texture analysis method, pre-process that picks up certain areas is proposed. Experimental results show that the comprehensive judgement between the two methods performs well to estimate the crowd density.
周宇男、王生进
电子技术应用
视频监控人群密度估计前景像素数统计视角归一化灰度共生矩阵
rowd monitoringEstimate crowd densitiesCompute the number of foreground elementsPerspective normalizationGrey level co-occurrence matrix
周宇男,王生进.基于融合判决的人群密度分析方法研究[EB/OL].(2013-01-10)[2025-08-16].http://www.paper.edu.cn/releasepaper/content/201301-484.点此复制
评论