基于完全局部二值模式的人群密度估计

Crowd density estimation based on completed local binary pattern

为了实时监控重要场所的人群密度、采取有效措施疏散高密度人群,避免人群密度过大而引发事故,造成生命和财产的损失,提出了一种基于完全局部二值模式的人群密度估计方法。该方法提取人群图像的3种局部纹理特征,建立了3-D联合直方图统计模型,用卡方距离最近邻方法对人群密度级别进行分类,实现了特定场景下人群密度的监测。对比实验结果表明了该方法能兼顾任意密度级别人群图像的分类,不仅准确率高,而且实时性好,同时对场景背景具有较强的鲁棒性。

To monitor the crowd density of a vital area and take efficient measures to disperse the high density crowd and avoid a series of accidents taking away lives and with great losses of properties, crowd density estimation based on a texture analysis method called ＂completed local binary pattern＂ is presented. Crowd images＇ local texture features are extracted and joint histogram models of each image are established, and the minimum chi-square distances of histograms are calculated and used to estimate the crowd density in a specified view. The approach presented can be applicable to any levels of crowd density images and get the higher rate of accuracy, the satisfying time consuming, and the strong background robustness, which are demonstrated by the results of the comvarative exoeriments.