基于GLCM的人群行为识别研究

随着社会经济的发展和城市化进程的加快,如何对公共场所人群行为进行监控,防止由于人群密度过高导致的重大伤亡事件发生是很有必要的。结合混合高斯模型和灰度共生矩阵提出一种监控人群行为的算法,将纹理特征通过灰度共生矩阵计算特征量,机器学习所有特征量来判断人群行为,并通过标准数据集和自行拍摄数据集验证算法的有效性。采用人群的整体特征来表征不同人群,通过机器学习来区分不同人群的行为情况,可用于安防监控、资源管理等领域。

A Method Based on GLCM for Crowd Behavior Recognition

With the development of social economy and urbanization, how to monitor the behavior of crowd in public places and prevent the fatalities of high density crowd accidents is necessary. In this paper, we propose a monitoring crowd behavior algorithm combining Gaussian Mixture Models and GLCM, and compute the texture feature using GLCM. Then we train all the features and judge the crowd behavior through machine learning, and use the standard data sets and validation data sets to verify the effectiveness of the proposed algorithm. The algorithm uses the whole crowd feature to express different kinds of crowd, then judges the behavior of different groups through machine learning. It can be used for security monitoring, resource management and other fields.