基于移动机器人图像融合的场景重建

发布时间：2018-06-29 15:02

本文选题：场景重建 + 全景图　；参考：《东华大学》2013年硕士论文

【摘要】：场景重建技术是计算机视觉的研究热点,随着计算机图形学技术的发展,以及城市规划、房地产、旅游、科教、军事等众多领域对虚拟仿真、展示需求的不断涌现,虚拟环境系统得到了越来越广泛的应用,并呈现出由单一绘制手段、对单一场景对象进行表现向多种绘制手段综合、对大规模、多对象复杂场景进行渲染发展的趋势。与此同时,现代信息感知、识别、传输和处理技术的不断提高,要求在虚拟环境系统中采用更为丰富和直观的手段对大量信息和对象进行建模和表现,为虚拟环境中信息的统一管理提供条件本文主要研究根据移动机器人摄像头获取的单张图像和多张图像信息实现单幅图片场景重建和多幅图像场景重建的方法。首先详细介绍了场景重建技术的研究背景与研究现状,从而展现其重要的研究意义与广阔的应用前景。然后对现有的场景重建方法进行介绍与比较,并确定使用全景图技术和灭线TIP技术来完成本次设计。本论文的主要工作成果如下： (1)构建基于图像几何变换的图像拼接理论。构建全景图建模整体理论,依现有的SIFT特征点提取、匹配与配准方法,利用图像几何变换关系自动求取理论解决图像的平移、图像的旋转以及不同尺度大小图像的拼接。 (2)灭点TIP图像建模计算方法研究。依托现有单幅照片图像建模方法,研究基于单灭点和双灭点的图像建模计算方法和重建的三维模型深度信息求取过程。 (3)构建灭线TIP图像建模的改进模型。分析基于灭点建模方法的缺陷产省原因构建灭点到灭线的过渡模型,提出了基于灭线建模的改进模型,并提出基于深度计算的图像漫游理论,实现灭线T工P模型漫游。 (4)全景图与灭点TIP两种图像建模方法的结合理论研究。研究两种方法在生成效果上存在互补性,实现全景图与灭点TIP的两个模型的平滑连接。
[Abstract]:Scene reconstruction technology is the research hotspot of computer vision. With the development of computer graphics technology and the demand for virtual simulation and display in many fields such as urban planning, real estate, tourism, science and education, military and so on, Virtual environment system has been more and more widely used, and it shows the trend that the single scene object is represented by multiple rendering methods, and the large-scale and multi-object complex scene is rendered and developed by a single rendering method. At the same time, the continuous improvement of modern information perception, recognition, transmission and processing technology requires the use of richer and more intuitive means to model and represent a large number of information and objects in the virtual environment system. This paper mainly studies the methods of single picture scene reconstruction and multiple image scene reconstruction based on single image and multiple image information obtained by mobile robot camera. Firstly, the research background and present situation of scene reconstruction technology are introduced in detail, so as to show its important research significance and broad application prospect. Then the existing scene reconstruction methods are introduced and compared, and the panoramic image technology and wire-out tip technology are used to complete the design. The main achievements of this thesis are as follows: (1) Image mosaic theory based on image geometric transformation is constructed. The whole theory of panoramic modeling is constructed. According to the existing sift feature points extraction, matching and registration, the theory of image geometric transformation is used to solve the problem of image translation automatically. The rotation of images and the stitching of images of different scales. (2) study on the modeling and calculation method of tip image with vanishing point. Based on the existing single photo image modeling methods, this paper studies the calculation method of image modeling based on single vanishing point and double vanishing point and the process of obtaining depth information of reconstructed 3D model. (3) an improved model of vanishing tip image modeling is constructed. Based on the vanishing point modeling method, the transition model from vanishing point to vanishing line is constructed, and an improved model based on vanishing line modeling is put forward, and the image roaming theory based on depth calculation is put forward. (4) A theoretical study on the combination of panoramic image and vanishing tip image modeling method. It is studied that the two methods are complementary to each other in generating effect and realize the smooth connection between the panorama and the two models of the vanishing tip.
【学位授予单位】：东华大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP242;TP391.41

【参考文献】