计算机科学

计算机科学 ›› 2023, Vol. 50 ›› Issue (6A): 211200205-7.doi: 10.11896/jsjkx.211200205

• 图像处理&多媒体技术 • 上一篇    下一篇

基于运动对比度增强的人群运动分割方法

张新峰, 倪启立, 陈舒涵, 杨宝庆, 李斌   

  1. 扬州大学信息工程学院(人工智能学院) 江苏 扬州 225127
  • 出版日期:2023-06-10 发布日期:2023-06-12
  • 通讯作者: 张新峰 (zhangxf@yzu.edu.cn)
  • 基金资助:
    国家自然科学基金(61801417,61802336);江苏省高等学校自然科学研究面上项目(18KJB520051)

Motion Contrast Enhancement-based Crowd Motion Segmentation Method

ZHANG Xinfeng, NI Qili, CHEN Shuhan, YANG Baoqing, LI Bin   

  1. College of Information Engineering(College of Artificial Intelligence),Yanzhou University,Yangzhou,Jiangsu 225127,China
  • Online:2023-06-10 Published:2023-06-12
  • About author:ZHANG Xinfeng,born in 1984,Ph.D,lecturer.His main research interests include visual computing and collaborative perception.
  • Supported by:
    National Natural Science Foundation of China(61801417,61802336) and Natural Science Foundation of the Jiangsu Higher Education Institutions of China(18KJB520051).

PDF (PC)

269

摘要： 公共场所监控视频中的人群运动状态复杂多变,很难通过检测或者跟踪每个个体来实现整个人群运动状态的分析,将人群分割成运动状态基本一致的区域成了了解和分析人群运动状态的有效途径。有监督的人群运动分割方法需要提供数据标注代价极高的像素级的训练集,因此无监督的聚类方法成为了更有前途的人群运动分割方法。然而,由于描述人群运动的局部特征通常是逐渐变化的,导致基于聚类的无监督方法需要针对不同的人群场景选择不同的参数,这很难适应各种不同的应用场景。为此,文中提出了一种基于运动对比度增强的人群运动分割方法。该方法是一种无监督模型,首先根据运动场中运动和噪声的分布特点增强不同运动状态之间的对比度,然后结合自适应阈值分割算法和标记符分水岭算法来提取每个运动状态基本一致的区域,避免了无监督聚类方法参数难以恰当选择的难题。在获得人群运动分割结果的基础上,文中提出了一种能量模型用于描述人群运动状态的稳定性。该能量模型通过推演出整个人群运动状态的变化过程来实现对异常人群运动状态的提前预警。在不同类型的复杂人群运动状态的场景中进行人群运动分割的实验,实验结果验证了基于运动对比度增强的人群运动分割方法的有效性和分割的准确性,以及所提能量模型的有效性。

关键词: 人群运动分析, 运动对比度增强, 人群运动分割, 能量模型, 异常预警

Abstract: In surveillance videos of public places,the movement states of the crowds are various and complex,and it is difficult to analyze the movement state of the whole crowd through detecting or segmenting every individual.Therefore,it is an effective way to understand and analyze the movement state of the crowd by dividing the crowd into areas with basically the same movement state.Supervised crowd motion segmentation methods require pixel-level training sets with high labeling costs,and thus unsupervised clustering methods are more promising for crowd motion segmentation.However,since the local features describing crowd movements usually change gradually,leading to the unsupervised methods based on clustering algorithm need to choose different parameters for different crowd scenarios,it is difficult to adapt to a variety of different application scenarios.To this end,this paper proposes a motion contrast improvement-based crowd motion segmentation method.The method is an unsupervised model that first enhances the contrast of different motion states based on the distribution law of movement and noise in the motion field,and then combines the adaptive threshold segmentation algorithm and the marker watershed algorithm to extract the essentially consistent region for each motion state,avoiding the difficulty of parameter selection for unsupervised clustering methods.Based on the results of crowd motion segmentation,this paper presents an energy model to describe the stability of crowd movement.The energy model can enable early warning of abnormal crowd motion state by deducing the change process of the whole crowd motion state.Experiments are conducted on crowd motion segmentation in different types of complex crowd motion state scenes.Experimental results verify the effectiveness and segmentation accuracy of the motion contrast enhancement-based crowd motion segmentation method and the validity of the proposed energy model.

Key words: Crowd motion analysis, Motion contrast enhancement, Crowd motion segmentation, Energy model, Abnormal warning

中图分类号: 

  • TP37
[1]ZENG Z M,WANG J.Research about Emergency Intelligence Information Service from the Social Computing Perspective[J].Journal of Intelligence,2017,36(11):59-64,77.
[2]SHARMA R,GUHA T.A trajectory clustering approach tocrowd flow segmentation in videos[C]//IEEE International Conference on Image Processing.IEEE,2016.
[3]CHEN M,WANG C S,HAO D H.Pedestrian Detection Algorithm Based on SSD Multi-modal and Multi-scale Feature Fusion[J].Journal of Jinling Institute of Technology,2021,37(2):33-38.
[4]LI J J,YANG H,WU S.Crowd semantic segmentation based on spatial-temporal dynamics[C]//2016 13th IEEE International Conference on Advanced Video and Signal Based Surveillance(AVSS).IEEE,2016.
[5]CHENG Y.Research on Crowd Segmentation Based on Rein-forcement Learning[D].Shanghai:Shanghai Jiao Tong University,2020.
[6]YU H,PAN G,ZHANG L,et al.Translation domain segmentation model based on improved cosine similarity for crowd motion segmentation[J].Journal of Electronic Imaging,2019,28(2):1.
[7]ZHANG X F,NI Q L,CHEN S H,et al.A Crowd Flow Segmentation Method based on Deep Motion Transformation Network[C]//2021 6th International Conference on Multimedia and Image Processing(ICMIP 2021).Zhuhai,China,2021:8-10.
[8]ZHANG L,HE Z W,GU M Y,et al.Crowd SegmentationMethod Based on Trajectory Tracking and Prior Knowledge Learning[J].Arabian Journal for Science & Engineering,2018,43:7143-7152.
[9]GE W,COLLINS R T,RUBACK R B.Vision-Based Analysis of Small Groups in Pedestrian Crowds[M].IEEE Computer Society,2012.
[10]JODOIN P M,BENEZETH Y,WANG Y.Meta-Tracking forVideo Scene Understanding[C]//2013 10th IEEE International Conference on Advanced Video and Signal Based Surveillance.IEEE,2013.
[11]RODRIGUEZ M,SIVIC J,LAPTEV L,et al.Data-driven crowdanalysis in videos[C]//2011 International Conference on Computer Vision.2011.
[12]NI Q L.Research on crowd flow segmentation method in complex scene and its application in crowd motion description[D].Yangzhou University,2021.
[13]ZHOU B,TANG X,WANG X.Coherent Filtering:Detecting Coherent Motions from Crowd Clutters[J].Springer Berlin Heidelberg,2012.
[14]PRAVEEN R G,BABU R V.Crowd flow segmentation based on motion vectors in H.264 compressed domain[C]//IEEE International Conference on Electronics.IEEE,2014.
[15]SUN D Q,YANG X D,LIU M Y,et al.PWC-Net:CNNs for Optical Flow Using Pyramid,Warping,and Cost Volume[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2018:8934-8943.
[16]CAO Y Q,WU D,HUANG X S.Track Defect Image Classification Based on Improved Ant Colony Algorithm[J].Computer Science,2019,46(8):292-297.
[17]GUO H R,SHAO W,ZHOU A W,et al.Novel defect recognition method based on adaptive global threshold for highlight metal surface[J].Chinese Journal of Scientific Instrument,2017,38(11):2797-2804.
[18]YU W S,HOU Z Q,SONG J J.Color Image SegmentationBased on Marked-Watershed and Region-Merger[J].Acta Electronica Sinica,2011,39(5):1007-1012.
[19]HAN J X.Applicable conditions of the law of conservation of momentum[J].Journal of Shenyang University(Social Science),2001,3(B12):211-212.
[20]WANG Z X.Law of conservation of energy[J].Physics Bulletin,1951(3):11-15.
[21]CHEN T,YING Z G,SHEN S F,et al.Evacuation simulation and analysis of social force model under the influence of relative speed[J].Progress in Natural Science,2006,16(12):1606-1612.
[22]WU H Q,XIE W H.Teaching of the First Law of Thermodynamics and the Second Law of Thermodynamics[J].The Guide of Science & Education,2015(4):80-81.
[23]ZHOU B L,TANG X O,ZHANG H P,et al.Measuring Crowd Collectiveness[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2014,36(8):1586-1599.
[24]SHAO J,LOY C C,WANG X G.Scene-Independent Group Profiling in Crowd[C]//Computer Vision and Pattern Recognition.IEEE,2014.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发