摘要
随着视频监控系统的大范围应用,提高监控系统的智能化程度势在必行。本文提出了一套基于运动物体检测的的智能视频监控方案,对部分关键技术作了深入研究,并实现了原型系统。
引入运动检测的目的是从视频序列的场景图像中检测出运动目标,以此来确定是否有物体入侵,并通过运动目标提取,进一步确定运动目标的细节,作为监控分析的依据,或作为犯罪证据保存下来。
本文提出了一种新的运动检测算法:在固定背景差分图像法的基础之上引入分块的思想;根据背景噪声高斯分布的特征确定了检测的阈值;加入了自适应亮度调节和参考帧更新机制。实验结果表明:相对于已有算法具有更好的性能,抗干扰能力更强,运动物体检测结果更准确、更可靠。本文采用基于块的顺序区域生长法进行运动目标提取,提出了采用区分规律性运动和非规律性运动的思路,以降低自然界中物体运动对检测效果的影响。
在此基础上,设计了一套完整的智能视频监控系统方案,并实现了可运行的原型系统。最终运行结果证明了设计的正确性和运动检测算法的有效性,整体效果良好。
With the application of video surveillance system widely, it's necessary to intelligentize surveillance system. This thesis brings forward an integrated design and solution of intelligent video surveillance system based on moving and change detection technologys, studys some key technologys of intelligent video surveillance system, and implement the archetypal system.
The purpose of introducing moving and change detection technologys is to detect moving objects. By picking up moving objects, we can get the details of moving objects, and save the video and picture as the evidence of crime.
This thesis brings forward a new algorithm of moving and change detection: The algorithm introduces partitioning method based on the frame difference algorithm with changeless image; decides limit range according to the character of gauss distributing of CCD noise; adds brightness adaptive adjustability and updating referenced image. Compared with other algorithm, it's more accurate and more credible.
This thesis designs an integrated solution of intelligent video surveillance system based on the moving and change detection algorithm, and implements exercisable archetypal system. The algorithm in this dissertation is verified by experiments and gets good results.
引文
[1]黎洪松.数字视频技术及其应用.北京:清华大学出版社,1998
[2]赵忠旭,沈兰荪,王爱民.视频图像采集系统的设计与实现.测控技术,1998(6)
[3]慧新标,郑志航.数字电视技术基础.北京:电子工业出版社,2005
[4]A.Murat Tekalp Digital Video Processing,北京:清华大学出版社,1998
[5]Kenneth.R.Castleman,Digital Image Processing,清华大学出版社,Prentice Hall
[6]Phlips.Data Sheet of SAA7111A.1998
[7]林志贤,廖志君,郭太良.SAA7111A在彩色视频FED中的应用.2005(51):48-52
[8]黄慧娟,王瑞光,丁铁夫.SAA7111A在平板显示视频接口中的应用.液晶与显示.2004.8:293-298
[9]Phlips.The I2C BUS Specification Version2.2000
[10]Samsung Electronics.S3C2410X 32-BIT MICROPROCESSOR USER'S MANUAL REVision 1.1.2003
[11]Collins R,Lipton A,Kanade T.Introduction to the Special Section on Video Surveillance[J].IEEE Trans.Pattern Analysis and Machine Intelligence,2000,22(8):745-746
[12]Maybank S,Tan T.Introduction-special Section on Visual Surveillance[J].International Journal of Computer Vision.2000,37(2):173
[13]阮秋琦.数字图像处理学.北京:电子工业出版社2001
[14](美)卡斯尔曼(Kenneth R.Castleman).数字图像处理.北京:电子工业出版社,2002
[15]章毓晋.图象工程(下册):图象理解与计算机视觉[M].北京:清华大学出版社,2000
[16]Horn B,Schunch B.Determining optical flow[J].Artificial Intelligence,1981.17:185-203
[17]Nagel H H.Constrains for the estimation of displacementvector fields from image sequences[A].In:Proceeding IJCAI'-83[C],Karlsruhe,Germany,1983:945-951
[18]Tretiak O,Pastor L.Velocity estimation from image sequences with second order differential operators[A].In:Proceedings Internation Conference on Pattern Recognition[C].Montreal,Canada,1984:16-19
[19]Otte M,Nagel H H.Estimation of optical flow based on higher-order spatiotemporal derivatives in interlaced and non-interlaced image sequences[J].Artificial Intelligence,1995,78(1/2):5-43
[20]Seferidis V.General approach to block-matching motion estimation[j].Optical Engineering,1993,32(1):9-79
[21]Tekalp AM.Digital video processing[M].New Jersey:Prentice-Hall inc,1996
[22]Marzani E,Legrand L,Dusserre L.Fourier adaptive smoothness constraint for computing optical flow on sequences of angiographic images[A].In:Proceedings of the IEEE Symposium on Computer-Based Medical Systems[C].Los Alamitos:1996:53-58
[23]Convertino G,Stella E,Branca A etal.Optic flow estimation by a Hopfield neural network using geometrical constraints[J].Pattern Recognition,1997,10(3):114-122
[24]Yasushi Mae,Yoshiaki Shirai,Jun Miura,etal.Object tracking in cluttered background based on optical flow and edges[EB/OL].http:// www-cv.mech.eng.osaka-u.ac.jp/research/tracking-group/mae/tlow-edge/ICPR96/ICP R96.html.
[25]Yasushi Mae,Shinya Yamamoto,Yoshiaki Shirai,etal.Optical flow based realtime object tracking by active vision System[EB/OL].http://www-cv.mech.eng.osaka-u.ac.jp/research/track2ing-group/mae/flow/JFCM94/JFCM94.html.
[26]Ryuzo Okada,Yoshiaki Shirai,Jun Miura.Object tracking based on optical flow and depth[EB/OL].http://www-cv.mech.eng.osaka-.ac.jp/research/tracking-group/okada/flow-disp-track-ing/flo w-disp-tracking.html.
[27]Ryuzo Okada,Shinya Yamamoto,Yasushi Mae.Realtime persontracking system[EB/OL],http://www-cv.mech.eng.osaka-u.ac.jp/research/tracking-group/okada/person-tracking/paper/TIcontest.ps.gz.
[28]谈新权.微光成象系统中实时降噪的优化设计.光电工程,1997,24(5):67-72
[29]M.Kim,J.G.Choi,D.Kim,M.H.Lee,C.Ahn and Y.S.Ho.A VOP Generation Tool:Automatic Segmentation of Moving objects in Image Sequences Based on Spatio Temporal Infornation.IEEE Trans.On Circuits and Systems for Video Technology,1999:1216-1226
[30]Snyder.DL.Compensation for readout noise in CCD image[J].J.Opt.Soc.Am.A,1995,12(2):272-83
[31]陈坚,李在铭.场景监视中的运动目标检测技术,信号处理,Vol(17)337-41
[32]徐建华.图像处理与分析.北京:科学出版社,1992
[33]Wik L and J,Grand G.Tracking of Multiple Moving Objects.In:Cappellinived Time Varying Image Processing and Moving Objects Recognition.[s.l.]:Eisevier Science Publishers,1987
[34]郑江滨,张艳宁,付亚奇,赵荣椿.视频监视中精确的目标提取算法.信号处理,Vol(17)342-345
[35]G.Hirzinger,K.Landzettel.A fast technique for segmentation and recognition of binary pattern.IEEE Computer Society Conference on Pattern Recognition and Image Processing.NY,USA:IEEE Comput Soc Press,1981.360-36