Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

Real-time video display system key technology research

Author RuanQin
Tutor PengGang
School Huazhong University of Science and Technology
Course Control Theory and Control Engineering
Keywords Video Image Stitching Feature Extraction Image Registration Seamless
CLC TP391.41
Type Master's thesis
Year 2011
Downloads 118
Quotes 0
Download Dissertation

Visual information to obtain the objective world of human information in the most direct way , so images, video, related processing technology in virtual reality, three-dimensional reconstruction of objects , video surveillance, video retrieval , medical image analysis and other fields is very important. Among them, the video image stitching technology can not reduce the resolution of cases, expand horizons, to eliminate redundancy in these application areas is even more widely used. In this paper, two ordinary USB camera for seamless video streaming and real-time display research, including camera calibration, distortion correction , feature extraction and matching, image fusion and video display and other parts of the research and analysis related technologies and comparison to achieve the desired effect. Firstly, studied a variety of calibration methods , using OpenCV library provides functions for camera calibration , in order to determine the internal and external parameters and distortion parameters , the image of the correction process . Then focus on the image registration techniques to compare the advantages and disadvantages of each method and select feature point based registration method . Meanwhile, also build descriptor Sift and Surf principles and procedures to give a detailed description and proved by experiments Surf Operators in algorithm complexity , speed, stability under a variety of transformation superior Sift operator . After extracting feature points in the image , use bbf feature alignment algorithm , combined ransac algorithm to calculate the image eliminate mismatch between the transformation matrix . Through the average method, the weighted average method , the experiment results , we propose a modified ranks of the center-weighted method for image stitching . Meanwhile , the use DirectShow technology to ensure the real-time video image stitching , to achieve a dual USB camera video stream seamless .

Related Dissertations
More Dissertations