Dissertation > Excellent graduate degree dissertation topics show

Video surveillance systems motion detection and tracking technology research

Author: ZhuBiZuo
Tutor: ZhengShiBao
School: Shanghai Jiaotong University
Course: Communication and information systems
Keywords: Gaussian mixture model Mean-Shift Particle filter Shadow Elimination Occlusion Handling
CLC: TP391.41
Type: Master's thesis
Year: 2009
Downloads: 138
Quote: 3
Read: Download Dissertation

Abstract


With the development of computer technology and artificial intelligence, machine vision technology research has made considerable progress, and attracted more and more attention. Currently, according to the complete machine vision motion detection, identification, tracking is widely used in national defense, aviation and marine, medical and health, safety monitoring and other fields of national economy. Motion detection and tracking is an important research direction. The purpose of motion detection is to detect whether there is a target within the monitored area and to determine the target object to appear, size and other information; target tracking target object for the selected trajectory description, related to the target feature extraction, feature descriptions, target matching and other aspects. The main objective of this study is to realize the simulation of human motion in video sequences target real-time monitoring. Detection and tracking system consists of two modules, respectively, the text detection and tracking algorithm analysis and research, and propose improvements and improved methods based on, and finally a combination of both to achieve a stationary background with faint dynamic background Automatic human target under quasi real-time tracking. This paper presents a dynamic fit stationary background with faint background video sequence moving target detection method. In this method, do the pixel Gaussian mixture modeling (Gaussian Mixture Model, referred GMM), in determining the pixel is a background or foreground points, the traditional GMM with a reference to the same location on the expectation and variance of pixel sample parameters strategy. However, the information between neighboring pixels of the pixel classification is essential. Based on this, in the decision process, it is proposed to do the former set of spatial domain attractions secondary judgment strategy. In a second judgment spatial judgment, based on the history of neighboring pixels using the statistical value information for heavy sentences can be screened out \issue. Another difficulty is the motion detector to eliminate shadows. In this paper, the background pixel brightness characteristics of the foreground pixels normalized quadratic spatial brightness judgment, also achieved good removal shadow effect. Finally, after a separation method for GMM background multiple test analysis, some parameters were optimized: the modeling space from one-dimensional to three dimensions, and redefines the variance convergence function. Experiments show that the spatial domain-based policy decision GMM secondary separation method can reduce the background pixel false detection rate and the effective suppression of the shadow moving objects. Tracking algorithm of this paper is mainly based on Mean-Shift algorithm and particle filter algorithm. In the Mean-Shift on the basis of theoretical analysis, first summarizes the Mean-Shift tracking algorithm applied on the concrete steps. Secondly, from the image processing point of view is defined on the target visibility, and gives the corresponding mathematical calculation. By target visibility and Mean-Shift effect of bandwidth of kernel function, elaborated in the target face cover and case tracking dynamic complex background of unsatisfactory. Cover a large area of ​​the target situation, this paper presents polynomial curve fitting method to estimate the movement is blocked when the target position; against the dynamic and complex background, this paper Particle Filtering and Mean-shift respective advantages, designed based on particle filter The Mean-shift tracking algorithm. Election of the image histogram tracking algorithm as the core target template features described in the tracking process can adaptively adjust the standard template, and can automatically adjust the size of the tracking area. Finally, on the PC platform implemented based EasyVideo digital video processing software development kit for moving object detection and tracking system. Experimental results show that the system works well for moving object detection, can effectively inhibit the shadows, and the target is obscured in the case of dynamic context and properly track the target.

Related Dissertations

  1. Research on Cooperative Orbit Determination in Satellite Network Based on Multi-Agent System Theory,V474
  2. The Maneuvering Target Tracking Research Based on VRPF,TN957.52
  3. The Research of 3D Human Motion Capture Based on Reference Points,TP391.41
  4. Research on the Theory of Visual Object Tracking Based on Particle Filter for Autonomous Robot,TP242
  5. Tracking Cells in High Density Image Sequences Based on Mean Shift Algorithm Combined with Topological Constraint,Q25
  6. Space Infrared Target Simulation and Application of Target Trace,TP391.41
  7. Research and Application on Detection and Tracking of Object Based on Video Stream,TP391.41
  8. Detection and Tracking of Moving Object in Complex Background,TP391.41
  9. Research on Object Detection and Tracking Method in Active Vision System,TP391.41
  10. Moving Objects Detection and Tracking Using Fisheye Camera,TP391.41
  11. The Research on the Target Localization and Tracking Based on WSN,TN929.5
  12. Key Algorithm in High Quality Voice Conversion System,TN912.3
  13. The Study of Moving Object Detection and Tracking Algorithm Based on Image Information,TP391.41
  14. Research of Moving Object Detection and Tracking Technology,TP391.41
  15. Object Detection and Tracking in Video Image,TP391.41
  16. Image Spam Detecting Based on Combinatorial and Statistical Classifier,TP391.41
  17. The Research on Localization and Target Tracking in Wireless Sensor Network,TN929.5
  18. Research of Image Segmentation in Web Image Search Based on GPUs,TP391.41
  19. AUV Integrated Navigation Algorithm Study and System Implementation,U666.1
  20. Architectural Optimizations for Particle Filters,TN713
  21. Video image sequence moving target acquisition and tracking,TP391.41

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device
© 2012 www.DissertationTopic.Net  Mobile