Dissertation > Excellent graduate degree dissertation topics show

The mobile robot based on three-dimensional laser ranging Scene Recognition

Author: LiYunHui
Tutor: ZhuangYan
School: Dalian University of Technology
Course: Control Theory and Control Engineering
Keywords: Three-dimensional laser data Azimuth Figure SURF Scene Recognition
CLC: TP242
Type: Master's thesis
Year: 2011
Downloads: 61
Quote: 0
Read: Download Dissertation

Abstract


Scene Recognition technology is the field of mobile robotics research important research direction, with the relocation of the mobile robot, maps maintenance, environmental reconstruction are closely related. For work in dynamic mobile robot indoor environment, due to collision or interference on the positioning failures are inevitable. At this point the robot current scene can be matched with the database scene, get the current scene pose to help robots to restore their position. This work is a three-dimensional laser range sensor system, we propose a combination of local and global spatial characteristics SURF feature scene identification algorithm. Scene Recognition process to overcome the body block, laser scanning posture deviation interference, this article will azimuth data into three-dimensional laser maps, and extract SURF feature as the local features of the scene. SURF features on translation, rotation, scale invariance with SIFT feature is similar, but the real-time feature extraction SURF significantly stronger than SIFT feature extraction. This paper studied the scene identification algorithm is proposed for the indoor scenes and indoor scenes with spatial area attributes and different scenarios, the space is usually also different. To be able to take advantage of differences in the area of ??the scene space to accelerate the scene identification process, this article from the scene of three-dimensional laser data extracted maximum scanning area of ??the scene as a scene of global features. Global features of the scene using scene recognition can significantly reduce the process to be identified in the scene and the scene database matched times, thereby increasing the scene identification algorithm in real time. For two match exact pose of the scene, we propose a matching function is used to evaluate the feature matching pairs calculated rotation translation matrix, and the scene that matches the highest degree of rotation and translation matrix is ??two scenes bits Pose relationship. In order to verify the effectiveness of the proposed algorithm and real-time, this paper chose one area of ??about 1600 square meters, with more than 40 rooms floor as an experimental site, and SmartROB2 mobile robot as the experimental platform. This paper identified the scene of the human body often occurs during static interference, human dynamic interference, laser ranging system scans pose interference testing. Interfere with the test results show that the characteristics of the Scene Recognition Based on SURF algorithm has strong robustness, and slightly stronger than the SIFT-based scene recognition algorithms. Test results show that the algorithm for real-time use of the proposed algorithm for two Scene Recognition takes about 300ms, about Scene Recognition Algorithm Based on SIFT half.

Related Dissertations

  1. Dose Colour Have Influence on Global Properties of Nature Scene Recognition?,TP391.41
  2. Studies of Technologies and Methods on Image Mosaic,TP391.41
  3. Research on Panorama Stitching Based on Speeded Up Robust Features,TP391.41
  4. Research on Video Forgery Passive Detection Based on Spatial-Temporal Feature,TP391.41
  5. Stereo Vision Based Tracking and Detection System for Serial Robot,TP242
  6. Outdoor target recognition and application in augmented reality,TP391.9
  7. SURF algorithm based on multiple images of three-dimensional model reconstruction method,TP391.41
  8. SURF algorithm based on improved electronic image stabilization technology,TP391.41
  9. SURF feature based on monocular vision SLAM Research and Implementation,TP242
  10. fMRI-based Brain Pattern Analysis Methods and Applications,TP391.41
  11. Missile seeker television image matching algorithm,TP391.41
  12. A class of visual perception and control of mobile robot technology research,TP242
  13. Research on Moving Target Tracking Technology in Video Images,TP391.41
  14. Based on SIFT scene understanding methods research,TP391.41
  15. The Study of Target Positioning Technology Based on UAV Image Sequence,TP391.41
  16. Panoramic Image Mosaic Based on SURF Feature Matchning Algorithm,TP391.41
  17. A Study on the Application of Characteristic Symbol Detection and the Scene Recognition Technology in Sport-video,TP391.41
  18. Research and Application on Invariant Feature Points of MMS Image Retrieval,TP391.41
  19. Study on Colored Image Registration Based on Feature,TP391.41
  20. Characteristic Comparison of Fish Larval and Juvenile Assemblages between the Surf Zones of South Branch of Yangtze River Estuary and North Coast of Hangzhou Bay,S931

CLC: > Industrial Technology > Automation technology,computer technology > Automation technology and equipment > Robotics > Robot
© 2012 www.DissertationTopic.Net  Mobile