Dissertation > Excellent graduate degree dissertation topics show

Vision-Based 3D Scene Modeling Research and Implementation

Author: LiWenKai
Tutor: ZhaoDeBin；YaoHongXun
School: Harbin Institute of Technology
Course: Computer Science and Technology
Keywords: Computer Vision 3D scene modeling Cluster adjustment Self-calibration Three-dimensional multi-view
CLC: TP391.41
Type: Master's thesis
Year: 2010
Downloads: 180
Quote: 0
Read: Download Dissertation

Abstract


With the development of information technology , more and more images and video in their lives and on the Internet . Moreover, the existing virtual reality technology and free viewpoint video technology are increasingly inclined to watch a digitized three-dimensional scene . Thus, the existing image and video organization to become a three-dimensional scene model , has become a very necessary and very promising work . Decades of rapid development of computer vision , visual method for three-dimensional scene modeling and laid a solid foundation. The three-dimensional scene modeling is an important research areas of the two disciplines of computer vision and computer graphics . Vision-based 3D scene modeling stage is to use computer vision methods , image or video processing , the final process of 3D scene model . Based graphics technologies , mainly as a post-processing to render and display the three-dimensional scene model . This article focuses on the study of vision-based three-dimensional scene modeling stage . This paper studies the work are as follows : First, this paper summarizes the important theory of the three - dimensional scene modeling and algorithms , Bundler self - calibration algorithm has been improved , greatly improving the efficiency of its iterative optimization . Second, this paper, and to achieve a robust two - view depth estimation method of three-dimensional scene lighting and equipment changes . Combining the SIFT features and NCC is not sensitive to light and equipment changes matching cost function : SIFT- _NCC , and second-order smoothness constraints MRF optimization algorithm stepwise refinement results than other type of algorithm is more precise. Third , this paper proposes the use of the unscaled image collection of large-scale three-dimensional scene modeling method . Develop effective image selection strategy , and get high-quality image subset to achieve optimization strategy of self - calibration method based cluster adjustment and MFF-MVS -based multi-feature fusion algorithm , using the scaling the collection of data and images reconstructed dense 3D point cloud data , and get a three-dimensional surface model .

Related Dissertations

  1. Moving target trajectory analysis based Intelligent Traffic Monitoring System,TP277
  2. A 60kA Dual-Loop Self-Balancing Direct Current Comparator,TM933.113
  3. Research of Cotton Nitrogen Integrated Management under Drip Irrigation and Plastic Film Mulch,S562
  4. Research on Siraidia Grosvenrii Features and Spezies Identification Based on Digital Image Processing and Artificial Neural Network,S567.239
  5. Design and Implementation of the Panoramic Virtual Tour System,TP391.41
  6. IC package oriented positioning system computer vision research,TP391.41
  7. Rectangularized CSS Images and a Matching Algorithm,TP391.41
  8. Research on Computer-Vision Based Motion Recognition Hidden Human-Computer Interface,TP391.41
  9. License plate location based on adaptive morphology and character segmentation algorithm research,TP391.41
  10. Based on machine learning methods of image edge detection and application,TP391.41
  11. Highway structured light vision inspection system calibration technology research,TP391.41
  12. Real-time tracking system for moving objects,TP391.41
  13. Car seat designed for the human body measurement system,TP274
  14. Based on computer vision technology, chlorophyll content detection system,TP274.4
  15. Bionic vision based UAV formation flying technology research,V249.1
  16. Research on On-line Detection System of Dehydrated Ginger by Computer Vision,TS255
  17. Real Time Object Recognition in Complex Environment,TP391.41
  18. Based on the three-dimensional image reconstruction algorithm sequence,TP391.41
  19. Computer vision-based multi-objective pose measurement method,TP29-AD
  20. Intelligent Video Analysis Technology for Elevator Cage Abnormal Behavior Detection Based on Computer Vision,TP274.4

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