静止背景中基于边界的运动目标检测技术研究
|
摘要
运动目标检测已成为图像处理技术与计算机视觉领域中的一个重要课题，运动目标检测多是基于底层视频信息的检测，由于大多数图像处理过程中考虑的是对应运动区域的像素，所以运动区域的有效提取对于目标分类、目标跟踪以及行为理解等工作具有重要的意义。运动目标检测有极其广泛的应用范围，比如：军事侦察、智能监控和交通检测等领域。用于运动目标检测的序列图像可分为静止背景、运动背景两种情况，本文是基于静止背景的研究。
文中首先概要介绍了运动目标检测的几种常用方法，包括光流法、帧间差分法、背景消减法，对各种方法进行了比较，并分析了其优缺点。在此基础上本文提出一种静止背景中基于边界的运动目标检测算法，本算法首先计算相邻两帧图像差分图像的阈值，将差分图像转化为二值图像，然后对二值图像进行形态学滤波确定运动目标的粗略位置，同时对当前帧进行边缘检测，将此边缘检测结果与前者确定的运动目标粗略位置进行交运算，得到运动目标的边界，再经滤波、区域填充等处理，最终获得运动目标的区域。
实验表明，本算法相对于传统的帧间差分法对运动目标速度要求低，较好的克服了其不能准确检测运动目标大小的问题，保持其对背景渐变不敏感的优点，本算法检测有效、稳定，无论单目标检测还是多目标检测，均取得了较好的效果。
Moving objects detection is one of the most important issues in image processing and computer vision. It belongs to bottom video information detection. Valid acquiring of moving region is important in object classifying, object tracking and behavior understanding because the pixels of the moving region are processed mainly. It is used widely in many fields, such as military reconnaissance, video surveillance and monitoring, traffic detection. Image sequences in moving objects detection can be divided into two types by which background is static or moving. The thesis is based on static background.
First, some common approaches in the field of detecting moving objects are introduced, including optical flow method, two consecutive frame subtraction method and background modeling method. Their advantages and disadvantages are compared and pointed out in this thesis. Then, an edge-based moving object detection algorithm is proposed. In the algorithm, the threshold of the difference images is calculated from the two consecutive frame images. Then the rough position of the moving object is decided by using morphological filter for binary image. Edge detection is carried for the current frame at the same time. For getting the edge of the moving object, the "and" operation is done between the results of rough position and the edge detection. At last, the area of the moving object is obtained through filtering and region filling.
Compared with traditional two consecutive from subtraction method, It solves the problem of not exactly detecting the size of moving objects, keeps the advantages of not sensitive for gradual changing background for the detection. The algorithm is valid, steady in object detection. It has obtained a better effect for not only single object detection but also multi-moving objects detection.
文中首先概要介绍了运动目标检测的几种常用方法，包括光流法、帧间差分法、背景消减法，对各种方法进行了比较，并分析了其优缺点。在此基础上本文提出一种静止背景中基于边界的运动目标检测算法，本算法首先计算相邻两帧图像差分图像的阈值，将差分图像转化为二值图像，然后对二值图像进行形态学滤波确定运动目标的粗略位置，同时对当前帧进行边缘检测，将此边缘检测结果与前者确定的运动目标粗略位置进行交运算，得到运动目标的边界，再经滤波、区域填充等处理，最终获得运动目标的区域。
实验表明，本算法相对于传统的帧间差分法对运动目标速度要求低，较好的克服了其不能准确检测运动目标大小的问题，保持其对背景渐变不敏感的优点，本算法检测有效、稳定，无论单目标检测还是多目标检测，均取得了较好的效果。
Moving objects detection is one of the most important issues in image processing and computer vision. It belongs to bottom video information detection. Valid acquiring of moving region is important in object classifying, object tracking and behavior understanding because the pixels of the moving region are processed mainly. It is used widely in many fields, such as military reconnaissance, video surveillance and monitoring, traffic detection. Image sequences in moving objects detection can be divided into two types by which background is static or moving. The thesis is based on static background.
First, some common approaches in the field of detecting moving objects are introduced, including optical flow method, two consecutive frame subtraction method and background modeling method. Their advantages and disadvantages are compared and pointed out in this thesis. Then, an edge-based moving object detection algorithm is proposed. In the algorithm, the threshold of the difference images is calculated from the two consecutive frame images. Then the rough position of the moving object is decided by using morphological filter for binary image. Edge detection is carried for the current frame at the same time. For getting the edge of the moving object, the "and" operation is done between the results of rough position and the edge detection. At last, the area of the moving object is obtained through filtering and region filling.
Compared with traditional two consecutive from subtraction method, It solves the problem of not exactly detecting the size of moving objects, keeps the advantages of not sensitive for gradual changing background for the detection. The algorithm is valid, steady in object detection. It has obtained a better effect for not only single object detection but also multi-moving objects detection.
引文
[1] 万缨．运动目标检测算法的探讨．计算机仿真．2006，23(10)：221-226页
[2] 王鲲鹏．序列图像运动目标检测算法研究．国防科技大学硕士学位论文．2005：1-3页
[3] 韩军，熊璋，李超，龚声蓉．分割视频运动对象的研究．计算机工程与应用．2000，8：22-26页
[4] 张远鹏等．计算机图像处理技术基础．北京：北京大学出版社，1996：94-96页
[5] 杨枝灵，王开．Visual C++数字图像获取、处理及实践应用．北京：人民邮电出版社，2003：156-176页
[6] J Barron,D Fleet,S Beauchemin.Performance of optical flow techniques[c]. International Journal of Computer Vision. 1994,12(1):42-77P
[7] 任荣新．基于运动图像的目标检测．中国海洋大学硕士学位论文．2006：11-12页，38-39页
[8] Yufei Ma,Hong-jiang,Zhang.Detecting.Motion Object by Spatio-temporal Entropy. IEEE International Conference on Multimedia and Expo(ICME2001).Waseda University, Tokyo,Japan,2001 (8):22-25P
[9] 马波，张田文．基于曲线进化的目标检测算法．信号处理．2003，19(3)：237-240页
[10] N Paragios,R Deriche.Geodesic active contours and level sets for detection and tracking of moving objects.pattern Analysis and Machine Intelligence.IEEE Transactions on 2000.22(3):266-280P
[11] Collins R et al. A system for Video Surveillance and Monitoring:VSAM final report. Carnegie Mellon University. Technical Report, CMU-RI-00-12.2000
[12] R Collins,A Lipton,and T kanade.Introduction to the special section on video surveillance[C].IEEE Trans Pattern Analysis and Machine Intelligence.August 2000,22(8):745-746P
[13] Ismail Haritaoglu ,Davlid Harwood,Larry S Davis.W4:A Real Time Surveillance of People and Their Activities[J].IEEE Transaction on Pattern Analysis and Machine Intelligence, 2000, 22(8):809-830P
[14] Wren R.Azarbayejani A,Darrell T, and Pentland A,Pfinder:real times for tracking of the human body[J] .IEEE Trans Pattem Anal Machine Intell. 1997,19(7):780-785P
[15] T.Olson,F.Brill.Moving Object Detection and Event Recognition Algorothms for Smart Cameras.Proc.DARPA Image Understanding Workshop, 1997,24(8): 159-175P
[16] 万琴．实时视频中多运动目标检测与跟踪研究．湖南大学硕士学位论文．2006：3-4页
[17] A J Lipton,H Fujiyoshi,and R S Patil.Moving target classification and tracking from real-time video[J].IEEE Workshop Applications of Computer Vision, 1998,8(17):8-14P
[18] 张文涛，李晓峰，李在铭．高速密集视频目标场景运动分析．电子学报．2000，28(10)：114-117页
[19] 章毓晋．图像工程(上册)图像处理．北京：清华大学出版社，2006：158-166页
[20] 张益贞．Visual c++实现MPEG／JPEG编码技术．北京：人民邮电出版社，2002：40-42页
[21] 胡小锋，赵辉．Visual C++／MATLAB图像处理与识别实用案例精选[M]．北京：人民邮电出版社，2004：67-115页
[22] 朱虹．数字图像处理基础．北京：北京科学出版社，2005：77-82页
[23] 朱秀昌，刘峰，胡栋．数字图像处理与图像通信．北京：北京邮电大学出版社，2002：77-78页
[24] 杨志刚．红外图像中机场的自动识别与跟踪．哈尔滨工程大学硕士学位论文．2007：8-9页
[25] 景晓军，周贤伟，付娅丽．图像处理技术及其应用．北京：国防工业出版社，2005：76-83页
[26] 李弼程，彭天强，彭波等．智能图像处理技术．北京：电子工业出版社，2004：149-160页
[27] Canny J.A computational approach to edge detection[J]. IEEE～PAMI. 1986,8(6):679-698P
[28] 贾静平．图像序列中多目标跟踪技术研究．西北工业大学硕士学位论文．2004：7页，12-13页
[29] B K P Horn,的 B G Schunch. Determinimg Optical Flow. AI 1981, (17):185-203P
[30] 徐建华．图像处理与分析[M]．北京：科学出版社，1992
[31] Horn B K P, Pobert Vision[M].Cambridge:Mit press, 1986: 56-60P
[32] 赵晓丽．基于统计降噪原理的动目标分割与跟踪．华北工学院硕士学位论文．2004：8-11页
[33] 牛少平．计算机视频监控技术研究．西北工业大学硕士学位论文．2006：12-13页
[34] 张杨．运动目标图像的检测与跟踪．中北大学硕士学位论文．2005：24-25页
[35] 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 Information.IEEE Yrans.OnCircuits and Systems for Video Technology. 1999.9(8):1216-1226P
[36] 朱辉．李在铭．视频序列中运动目标检测技术．信号处理．2002，18(5)：449-451页
[37] 唐波，孙茂印，汤晓安．基于自适应噪声滤波的视频对象分割算法．计算机应用研究．2007，(2)：173-175页
[38] 沈海浪．视频序列中运动目标的检测与跟踪．解放军信息工程大学硕士学位论文．2004：29-30页
[39] 贾永红．计算机图像处理与分析[M]．武汉：武汉大学出版社，2001
[40] 赵艳启．运动目标识别与跟踪算法的研究．河海大学硕士学位论文．2005：27-30页
[41] 李刚，邱尚斌，林凌，曾锐利．基于背景差法和帧间差法的运动目标检测方法．仪器仪表学报．2006，27(8)：961-963页
[42] 李瑛．复杂背景下动目标图像跟踪方法的研究．中北大学硕士学位论文．2006：12-13页，24-25页
[2] 王鲲鹏．序列图像运动目标检测算法研究．国防科技大学硕士学位论文．2005：1-3页
[3] 韩军，熊璋，李超，龚声蓉．分割视频运动对象的研究．计算机工程与应用．2000，8：22-26页
[4] 张远鹏等．计算机图像处理技术基础．北京：北京大学出版社，1996：94-96页
[5] 杨枝灵，王开．Visual C++数字图像获取、处理及实践应用．北京：人民邮电出版社，2003：156-176页
[6] J Barron,D Fleet,S Beauchemin.Performance of optical flow techniques[c]. International Journal of Computer Vision. 1994,12(1):42-77P
[7] 任荣新．基于运动图像的目标检测．中国海洋大学硕士学位论文．2006：11-12页，38-39页
[8] Yufei Ma,Hong-jiang,Zhang.Detecting.Motion Object by Spatio-temporal Entropy. IEEE International Conference on Multimedia and Expo(ICME2001).Waseda University, Tokyo,Japan,2001 (8):22-25P
[9] 马波，张田文．基于曲线进化的目标检测算法．信号处理．2003，19(3)：237-240页
[10] N Paragios,R Deriche.Geodesic active contours and level sets for detection and tracking of moving objects.pattern Analysis and Machine Intelligence.IEEE Transactions on 2000.22(3):266-280P
[11] Collins R et al. A system for Video Surveillance and Monitoring:VSAM final report. Carnegie Mellon University. Technical Report, CMU-RI-00-12.2000
[12] R Collins,A Lipton,and T kanade.Introduction to the special section on video surveillance[C].IEEE Trans Pattern Analysis and Machine Intelligence.August 2000,22(8):745-746P
[13] Ismail Haritaoglu ,Davlid Harwood,Larry S Davis.W4:A Real Time Surveillance of People and Their Activities[J].IEEE Transaction on Pattern Analysis and Machine Intelligence, 2000, 22(8):809-830P
[14] Wren R.Azarbayejani A,Darrell T, and Pentland A,Pfinder:real times for tracking of the human body[J] .IEEE Trans Pattem Anal Machine Intell. 1997,19(7):780-785P
[15] T.Olson,F.Brill.Moving Object Detection and Event Recognition Algorothms for Smart Cameras.Proc.DARPA Image Understanding Workshop, 1997,24(8): 159-175P
[16] 万琴．实时视频中多运动目标检测与跟踪研究．湖南大学硕士学位论文．2006：3-4页
[17] A J Lipton,H Fujiyoshi,and R S Patil.Moving target classification and tracking from real-time video[J].IEEE Workshop Applications of Computer Vision, 1998,8(17):8-14P
[18] 张文涛，李晓峰，李在铭．高速密集视频目标场景运动分析．电子学报．2000，28(10)：114-117页
[19] 章毓晋．图像工程(上册)图像处理．北京：清华大学出版社，2006：158-166页
[20] 张益贞．Visual c++实现MPEG／JPEG编码技术．北京：人民邮电出版社，2002：40-42页
[21] 胡小锋，赵辉．Visual C++／MATLAB图像处理与识别实用案例精选[M]．北京：人民邮电出版社，2004：67-115页
[22] 朱虹．数字图像处理基础．北京：北京科学出版社，2005：77-82页
[23] 朱秀昌，刘峰，胡栋．数字图像处理与图像通信．北京：北京邮电大学出版社，2002：77-78页
[24] 杨志刚．红外图像中机场的自动识别与跟踪．哈尔滨工程大学硕士学位论文．2007：8-9页
[25] 景晓军，周贤伟，付娅丽．图像处理技术及其应用．北京：国防工业出版社，2005：76-83页
[26] 李弼程，彭天强，彭波等．智能图像处理技术．北京：电子工业出版社，2004：149-160页
[27] Canny J.A computational approach to edge detection[J]. IEEE～PAMI. 1986,8(6):679-698P
[28] 贾静平．图像序列中多目标跟踪技术研究．西北工业大学硕士学位论文．2004：7页，12-13页
[29] B K P Horn,的 B G Schunch. Determinimg Optical Flow. AI 1981, (17):185-203P
[30] 徐建华．图像处理与分析[M]．北京：科学出版社，1992
[31] Horn B K P, Pobert Vision[M].Cambridge:Mit press, 1986: 56-60P
[32] 赵晓丽．基于统计降噪原理的动目标分割与跟踪．华北工学院硕士学位论文．2004：8-11页
[33] 牛少平．计算机视频监控技术研究．西北工业大学硕士学位论文．2006：12-13页
[34] 张杨．运动目标图像的检测与跟踪．中北大学硕士学位论文．2005：24-25页
[35] 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 Information.IEEE Yrans.OnCircuits and Systems for Video Technology. 1999.9(8):1216-1226P
[36] 朱辉．李在铭．视频序列中运动目标检测技术．信号处理．2002，18(5)：449-451页
[37] 唐波，孙茂印，汤晓安．基于自适应噪声滤波的视频对象分割算法．计算机应用研究．2007，(2)：173-175页
[38] 沈海浪．视频序列中运动目标的检测与跟踪．解放军信息工程大学硕士学位论文．2004：29-30页
[39] 贾永红．计算机图像处理与分析[M]．武汉：武汉大学出版社，2001
[40] 赵艳启．运动目标识别与跟踪算法的研究．河海大学硕士学位论文．2005：27-30页
[41] 李刚，邱尚斌，林凌，曾锐利．基于背景差法和帧间差法的运动目标检测方法．仪器仪表学报．2006，27(8)：961-963页
[42] 李瑛．复杂背景下动目标图像跟踪方法的研究．中北大学硕士学位论文．2006：12-13页，24-25页