Font Size: a A A

Research On Several Key Techniques Of 3D Panoramic Image Generation

Posted on:2017-01-04Degree:MasterType:Thesis
Country:ChinaCandidate:Y ZhaoFull Text:PDF
GTID:2348330488466034Subject:Signal and Information Processing
Abstract/Summary:
In this paper studies the techniques of 3D panoramic image generation can solve that one 3D-camera’s angle is too small and can not shoot large field of view.Its emergence has brought a far-reaching impact on human’s life,study and work.It has been used in more domains.Such as construction,medical,military and so on.The techniques of 3D panoramic image generation can be broadly divided into three categories.The first one is using the existing professional software to reconstruct the 3D panoramic image,the second one is that reconstruct the 3D panoramic image based on image,the last one is that get the 3D panoramic image through three-dimensional scanning equipment.This paper mainly studies the second and third technologies of3 D panoramic image generation.In this paper first use 3D scanning equipment SR-4000 3D camera to obtain the 3D point cloud data of the scenery which want to generate 3D panoramic image,then doing point cloud treatment,extracting point cloud’s feature,aligning point cloud,matching point cloud in order to generate 3D panoramic image with wide perspective.During the research found that when using SR-4000 3D camera to collect the point cloud outdoors,the precision of the point cloud is greatly influenced by environmental illumination and can not achieve accurate and complete point cloud data,and then can not doing point cloud pre-processing,point cloud alignment,point cloud matching to generate 3D panoramic image.For removing the above scheme’s shortcoming.In this paper put forward another kind of scheme that is the use of binocular vision to achieve the 3D panoramicimages.First,extracting feature,correcting,stereo matching and 3D reconstruction on the a set of images acquired by using binocular vision system to get the point cloud data.Second,rotating the binocular vision platform at a certain angle to collect second sets of images to generate point cloud data.At last,doing treatment,extracting feature,aligning,matching on the point cloud data which obtained from two adjacent moments to generate 3D panoramic image.After comparing the 3D panoramicimage generated by the above two methods,can conclude that using of binocular vision to achieve the 3D panoramic images can get more accurate and complete point cloud and has higher cost performance than using SR-4000 3D camera to generate the 3D panoramic image.
Keywords/Search Tags:Binocular Vision, Three-dimensional reconstruction, Point cloud processing, Point cloud alignment, Point cloud matching
Related items