CAJ | 학술논문

公共场景监控下的人群人数统计是公共安全管理中的一个重要内容。针对复杂场景监控的情况，本文提出一种融合像素与纹理特征的人群人数统计方法。首先，通过改进的视觉背景提取方法得到高精度的前景目标；然后，提取 ROI区域前景像素统计特征与纹理特征并引入阈值判别机制；最后，对阈值上下的图像分别采用基于纹理特征的回归算法和基于像素统计特征的线性拟合算法来得到人群人数。实验结果表明，本文方法相较于传统算法，可以适应于不同密度场景下的人群人数统计，且计算简单，统计精度高。
공공장경감공하적인군인수통계시공공안전관리중적일개중요내용。침대복잡장경감공적정황，본문제출일충융합상소여문리특정적인군인수통계방법。수선，통과개진적시각배경제취방법득도고정도적전경목표；연후，제취 ROI구역전경상소통계특정여문리특정병인입역치판별궤제；최후，대역치상하적도상분별채용기우문리특정적회귀산법화기우상소통계특정적선성의합산법래득도인군인수。실험결과표명，본문방법상교우전통산법，가이괄응우불동밀도장경하적인군인수통계，차계산간단，통계정도고。
The crowd counting under the public scene monitoring is an important part of public security management.For the case of complex scenes monitoring,this paper suggests a kind of fusion methods of pixels and texture feature for crowd counting.Firstly,the algorithm obtain a highly accurate foreground obj ect by the improved visual background extraction;and then,it extracts the ROI foreground pixels statistical characteristics and texture features and introduces a thresh-old determination mechanism;finally,images of the upper and lower thresholds are based on tex-ture features with regression algorithm and the statistical characteristics of the pixel fitting algo-rithm to get the number of the crowd.The experiment results show that the algorithm in this paper has the characteristic of a simple calculation and high statistical accuracy,which can be adapted to the people number statistics at different density scene compared with the traditional methods.