醫院監控場景下的人群密度估計方法

禹明娟,張英烈,陳臨強

(杭州電子科技大學 計算機學院,浙江 杭州　310018)

?

醫院監控場景下的人群密度估計方法

禹明娟,張英烈,陳臨強

(杭州電子科技大學 計算機學院,浙江 杭州310018)

摘要人群密度估計是智能化人群監控中的重要內容,在公共安防、管理控制和商業決策等方面起著重要作用。文中針對醫院應用場景,采用一種基于分塊的方法,對每一個子圖像分別利用基于像素特征與最小二乘直線擬合方法進行人數定量分析和基于灰度共生矩陣與支持向量機的方法進行密度定性分析,得到整幅圖像中不同子圖及整幅圖像的人數和密度分布圖。實驗表明,該方法能有效的提高人群密度估計的準確率,且還能對局部的密度異常精準定位。

關鍵詞人群密度估計;醫院;最小二乘法;灰度共生矩陣;支持向量機

Crowd Density Estimation Method for Hospital Surveillance

YU Mingjuan,ZHANG Yinglie,CHEN Linqiang

(School of Computer Science,Hangzhou Dianzi University,Hanghzhou 310018,China)

AbstractCrowd density estimation,with increasing attention,is the primary content of intelligent crowd surveillance.It plays an important role in the public security,management control as well as business decision.In this paper,we apply it in the situation of hospital with partition methods.We firstly divide the crowd image to sub images.Then for every sub image,we conduct quantitative analysis to the number of people with the function based on pixel feature and least-square line regression respectively.We also conduct a qualitative analysis of the density of people with the function based on gray level co-occurrence matrix (GLCM) and support vector machine (SVM).The number and density distribution of different sub images included in the whole image are obtained for real-time monitoring of the number of people in hospital with accurate location of the local density abnormity.

Keywordscrowd density estimation;the situation of hospital;least squares method;GLCM;SVM

近年來,隨著我國醫院就診人數的不斷增長,各類大型綜合性醫院一號難求、人滿為患的現象突出,對醫院的就醫環境和醫療秩序產生隱患[1]。因此,結合先進的圖像處理和機器學習技術,對醫院內的人群密度進行實時監測,對密度異常區域進行及時預警并及時采取措施疏散人群,合理分配醫療服務資源,維護就醫秩序,保障就醫環境是必要的。

根據人群密度特征的表征方式不同,人群密度估計方法[2]通常分為基于像素統計、基于紋理分析和基于目標分析3類方法。文獻[3]提出基于像素統計方法,而文獻[4]改進了這種方法,引入神經網絡估計人群密度?！?br>