Dissertation > Excellent graduate degree dissertation topics show

Fast Object Detection, Positioning and Motion Analysis

Author: SongJian
Tutor: LiuYunCai
School: Shanghai Jiaotong University
Course: Control Theory and Control Engineering
Keywords: object detection dominant orientation templates 3D human motion analysis annealed particle filter Markov random field parallel com-puting GPU OpenCL
CLC: TP391.41
Type: Master's thesis
Year: 2012
Downloads: 138
Quote: 1
Read: Download Dissertation

Abstract


Vision system of human beings is quite precise and complicated. People are trying to copy such a system with the function of perceive to the comput-er with the help of booming computer science and artificial intelligence, and then born the computer vision. The root task of computer vision is to perceive, and both visual information based object detection and motion analysis are the basis for intelligent systems to understand the world. In industry, more and more image sensors are taking place of the ordinary ones as they can catch huge information that no other can, urging a speedup of image detec-tion technology. And visual information based 3D human reconstruction and motion tracking is growing up in the field of virtual reality, film animation, intelligent surveillance, sports and medical treatment.Visual information is usually expressed in two or three or higher dimen-sions, and the huge data inside catalyzed both information explosion and the difficulty to analysis data in real time. Then comes the high performance computing, with its technology of speedup, and more recently GPU compu-ting and OpenCL, into the computer vision field.This thesis is established in designing a real-time object detection and motion analysis system. It first review recent researches about object detec-tion and human motion analysis, and then implements real-time detection al- gorithm for industry products and real-time human reconstruction and track-ing algorithms with the help of GPU and other parallel computing methods. And with those algorithms, this thesis established a real-time object detection and motion analysis system. The major research contents and results of this thesis are as follows.1) Proposed a high speed and template based object detection approach. It can detect multiple texture-less objects from the same scene in less than 100 ms using a coarse to fine search. This approach is designed for industry conditions.2) Proposed an approach combining human reconstruction and tracking under sparse views. It learns the state space online of the tracking algorithm and feed back tracking results to reconstruction approach and improves both.3) Accelerated human reconstruction and tracking algorithms using GPU, leading them to real-time. It accelerated the algorithms using multi-core CPU and AMD GPU, achieving a speedup of 400 and speed of 10 FPS.4) Established a real-time computer vision system.The well designed software system is coded in C++, brings real-time computing and system engineering to computer vision with the help of high quality capture and store devices. The part with GPU algorithms of the system has won the championship in the 2011 AMD China accelerated computing contest.

Related Dissertations

  1. The Research of Moving Object Tracking System Based on Embeded Image Process Unit,TP391.41
  2. Research on Visual Detection and Tracking of Mobile Robots,TP242.62
  3. Based on statistical methods of magnetic resonance imaging of the human brain image segmentation and three-dimensional analysis of the data,R445.2
  4. Visual Feedback and Memory Behavior Based GPU Parallel Ant Colony Algorithm,TP301.6
  5. The Airborne LiDAR and LiDAR Points-cloud’s Quick Processing Method,TN959.73
  6. Study on Visual Target Detection Based on SVM,TP391.41
  7. Research on the Algorithms of Moving Object Detection and Tracking,TP391.41
  8. The Research on Feature Selection for Data Stream,TP311.13
  9. Research on Medical Image Segmentation Method Based on Markov Random Field Model,TP391.41
  10. Research on Techniques in Video Surveillance,TP391.41
  11. Research of Sub-Diffraction in Optical Diffraction Field,O436.1
  12. The Research and Implementation of Background Modeling and Updating Algorithm Based on Mixture Gaussian Model,TP391.41
  13. Research on Object Detection and Tracking Method in Active Vision System,TP391.41
  14. The Anisotropic Etching Simulation of Silicon Based on Cellular Automata Model,TB304
  15. Research of Pedestrian tection Method Based on Active Contour Models,TP391.41
  16. Application of Graphics Processing Unit in Matrix Inversion and Normal Mode Analysis,TP391.41
  17. Recognition of Multi-class Objects Based on Intra-category Shared Features Model in Remote Sensing Images,TP751
  18. Research and Design of a High-Performance Scalable Public Key Cryptographic Coprocessor,TN918.1
  19. Research on Video Compression Algorithm Based on Multi-core Computing Platform,TN919.81
  20. GPU-Based Parallel Optimization of Adaptive Gaussian Mixture Background Modeling Algorithms,TP391.41
  21. Research of Finite Element Method on GPU,O241.82

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