多視點(diǎn)圖像的三維重建系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)
[Abstract]:Image-based 3D reconstruction is a reverse engineering which combines computer graphics, computer vision, pattern recognition and so on. This technology has a wide range of applications in the production of social life, including robot vision, medical imaging, virtual display, life entertainment and so on. Three-dimensional reconstruction technology is the process of forming three-dimensional structure of original scene from two-dimensional image obtained from ordinary camera. In this process, the camera's position and projection parameters are first calculated by feature point extraction and matching, then the dense point cloud of 3D scene is obtained by matching dense points of polar geometry, and then the dense point cloud is gridded. The 3D polyhedron model of the target scene is reconstructed. Finally, the original image is mapped to the polyhedral model to obtain a realistic 3D texture model. In order to realize the 3D reconstruction system based on image, it is necessary to master the basic theory of 3D reconstruction such as single hole camera imaging model, camera calibration, polar geometry and so on. In the aspect of feature extraction and matching, classical scale invariant feature change (SIFT) algorithm and normalized cross-correlation optimization algorithm are used, and RANSAC algorithm is used to optimize feature matching, to delete wrong matching points and to improve the accuracy of matching. In the process of dense point generation, the block-based dense algorithm (PMVS),) is used to deal with the scene with incomplete texture, and Power Crust algorithm is used to triangulate the point cloud. In the final texture processing process, we use the open source polyhedron processing tool MeshLab to map the original image to the surface of the 3D model to obtain the final solid model. Finally, the results of the three-dimensional reconstruction system are quantitatively analyzed by comparing with the results obtained by the laser scanner. The results show that the accuracy of the three-dimensional reconstruction system is up to the level of application.
【學(xué)位授予單位】:內(nèi)蒙古大學(xué)
【學(xué)位級(jí)別】:碩士
【學(xué)位授予年份】:2017
【分類號(hào)】:TP391.41
【參考文獻(xiàn)】
相關(guān)期刊論文 前7條
1 陳樹(shù);王磊;;一種改進(jìn)的基于RANSAC方法的SIFT特征匹配[J];信息技術(shù);2016年12期
2 張艷;李強(qiáng);;基于逐點(diǎn)插入法生成Voronoi圖的算法研究及實(shí)現(xiàn)[J];黑龍江工程學(xué)院學(xué)報(bào);2016年05期
3 謝利民;黃心漢;宋展;;基于不同光照條件的三維重建算法[J];華中科技大學(xué)學(xué)報(bào)(自然科學(xué)版);2013年S1期
4 Nguyen Tien Thanh;劉修國(guó);王紅平;于明旭;周文浩;;基于激光掃描技術(shù)的三維模型重建[J];激光與光電子學(xué)進(jìn)展;2011年08期
5 劉雪娜;;三維點(diǎn)集Voronoi圖的算法實(shí)現(xiàn)[J];計(jì)算機(jī)輔助工程;2006年01期
6 孟曉橋,胡占義;攝像機(jī)自標(biāo)定方法的研究與進(jìn)展[J];自動(dòng)化學(xué)報(bào);2003年01期
7 丁永祥,夏巨諶,,王英,肖景容;任意多邊形的Delaunay三角剖分[J];計(jì)算機(jī)學(xué)報(bào);1994年04期
相關(guān)博士學(xué)位論文 前2條
1 張學(xué)賀;基于雙目視覺(jué)的六足機(jī)器人環(huán)境地圖構(gòu)建及運(yùn)動(dòng)規(guī)劃研究[D];哈爾濱工業(yè)大學(xué);2016年
2 康來(lái);基于圖像點(diǎn)特征的三維重建方法研究[D];國(guó)防科學(xué)技術(shù)大學(xué);2013年
相關(guān)碩士學(xué)位論文 前3條
1 張也;基于局部降維的帶約束三維Delaunay三角網(wǎng)構(gòu)建算法研究[D];東南大學(xué);2016年
2 戴嘉境;基于多幅圖像的三維重建理論及算法研究[D];上海交通大學(xué);2012年
3 李競(jìng)超;基于立體視覺(jué)的三維重建[D];北京交通大學(xué);2010年
本文編號(hào):2463908
本文鏈接:http://sikaile.net/shoufeilunwen/xixikjs/2463908.html