基于灰度共生矩阵和分形的人群密度估计

针对人群密集公共场所的视频监控,传统的人工监控因为其局限性,已不能满足实际需要,人群智能监控应运而生,而人群密度成为监控的重要对象。基于像素点统计的人群密度估计方法简单直观,但仅适用于人群密度较低场合,密度较高时误差较大。对中高密度人群,本文给出了一种基于灰度共生矩阵和分形的人群特征提取方法,进而利用支持向量机实现人群密度分类。对基于视频的人群密度估计实验结果表明本文提出的方法是有效的。

Crowd density estimation based on grey co-occurrence matrix and fractal

In the light of video monitoring in the public place of crowd density,because of its limitations the traditional manual control could not meet the actual needs,so the crowd intelligent monitoring emerges as the times require and crowd density becomes the important object of video monitoring.The pixel-counting based crowd estimation is simple but the estimated error increases as the population density increases.This paper proposes a novel crowd density estimation method to higher density images,which first employs grey cooccurrence matrix and ffactal to extract the features,then uses the support vector machine to identify the level of the crowd density.According to the experiments of crowd density estimation based on video,the results show that this method is effective.