›› 2015, Vol. 28 ›› Issue (7): 129-.

• 论文 • 上一篇    下一篇

基于像素统计和纹理特征的人群密度估计

王强,孙红   

  1. (1.上海理工大学 光电信息与计算机工程学院,上海 200093;2.上海现代光学系统重点实验室,上海 200093)
  • 出版日期:2015-07-15 发布日期:2015-07-13
  • 作者简介:王强(1989—),男,硕士研究生。研究方向:模式识别与智能系统,人群密度估计。E-mail:305702058@qq.com。孙红(1964—),女,副教授。研究方向:计算机网络通信与云计算,计算机科学与技术,控制科学与工程。
  • 基金资助:

    国家自然科学基金资助项目(61170277,61472256);上海市教委科研创新重点基金资助项目(12zz137);沪江基金资助项目(C14003)

Crowd Density Estimation Based on Pixel and Texture

WANG Qiang,SUN Hong   

  1. (1.School of Optical-Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China;
    2.Shanghai Key Lab of Modern Optical System,Shanghai 200093,China)
  • Online:2015-07-15 Published:2015-07-13

摘要:

提出了一种人群密度估计算法,将像素统计和纹理特征两种基本方法进行有效结合。前景提取使用改进的Vibe算法,设定感兴趣区域(ROI)来减少运算量。同时,引入形态学处理和透视矫正消除了因人物远近所造成的误差。并设定了一套人群密度等级划分的标准,克服了因人群密度高低频繁变化造成的误差。最终,实验结果显示运算速度和正确率均较为可观,证明了本算法的可靠性。

关键词: 像素统计, 纹理特征, 前景提取, 人群密度

Abstract:

A crowd density estimation algorithm based on the pixel and texture is proposed.The Vibe algorithm is adopted for foreground extraction with the interested area (ROI) set to reduce the computational complexity.To solve the errors caused by the distance of the crowd,the morphology processing and perspective correction is involved.A standard is introduced to divide the crowd density to overcome the errors induced by changes in the crowd density.The experimental results show that the arithmetic speed and the accuracy are considerable,which verifies the reliability of the algorithm.

Key words: pixel;texture;foreground extraction;crowd density

中图分类号: 

  • TP391.41