Font Size: a A A

Research On Panoramic Roaming And Scene Matching Based On Fisheye Lens

Posted on:2018-06-18Degree:MasterType:Thesis
Country:ChinaCandidate:F WengFull Text:PDF
GTID:2428330572465547Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
Panoramic roaming technology is the use of panoramic image representation of virtual environment virtual reality technology,the technology allows people to quickly browse never been to the scene,and can give users a 360-degree viewing angle of the omni-directional,The feeling of its territory.Panoramic roaming technology uses image-based rendering method,compared with the traditional three-dimensional modeling technology,panoramic roaming technology to achieve simple,strong sense of reality,the system performance requirements.It has these advantages so that it has now been widely used,such as the major companies Street View,real estate developers,real estate preview and scenic spots of the scenic spots,such as preview.Visual navigation is superior to traditional navigation,it can work in the GPS can not work environment,such as the indoor environment,high-rise building environment;it's much lower cost,and than the traditional location contains richer visual information.As the important component of vision,scene matching is widely used in navigation system,missile guidance,remote sensing and pattern recognition,so the research of scene matching becomes the hot point of visual navigation.In order to realize the system,the main research contents are as follows:(1)Panoramic camera is used to collect a large number of natural scene images on the campus scale,and the fisheye images are calibrated and stitched together,and a panoramic image library of the campus,including source images and test images,is built.Aiming at the problem of registration in image mosaic,an image registration method based on block matching is proposed.Image registration is carried out based on block matching,which greatly improves the efficiency of registration.(2)Combining the characteristics of ORB and SIFT features,this paper proposes an algorithm based on ORB and SIFT for scene matching.After comparing and analyzing the experimental results,it verifies the correctness and efficiency of the matching algorithm,and improves the system running speed.(3)In order to apply the camera and people's perspective,this paper introduces the texture mapping to transform the cylinder panorama into the cube panorama.The introduction of the texture mapping greatly reduces the computation.The cube panorama can be produced very quickly,and the obtained image is also very Clear.(4)A panoramic browsing system based on HTML5 is designed and developed,which can make the users browse the panorama images freely on the webpage,and it can be controlled by the mouse,which has certain application value.Based on the realization of panoramic roaming,-this-paper develops scene matching technology,and realizes the fast query of image based on feature,and it can detect out accurately.With the most similar panoramic images,user-friendly panoramic view of the image in time.
Keywords/Search Tags:image correction, panoramic image, panoramic roaming, SIFT feature, scene matching
PDF Full Text Request
Related items