本文選題：多源測量數(shù)據(jù) + 多視匹配　； 參考：《中國地質大學》2017年博士論文

【摘要】：運用傾斜攝影技術是當前無人機航測的常用有效方法。傾斜攝影技術可以實現(xiàn)多角度影像采集,保證高效率、增加精確度、準確地理定位,實現(xiàn)三維實景重建,已成為國內(nèi)外無人機航測的主要技術途徑,并逐漸應用于地球物理探測等領域。測繪技術是實現(xiàn)土地利用規(guī)劃的基礎和關鍵。土地整治規(guī)劃目前主要采用全野外人工測量和傳統(tǒng)二維圖形設計,存在野外工作量大、成圖效率低、運行成本高等缺點,無人機航測成為解決上述問題的有效途徑,研發(fā)適用于無人機傾斜攝影測量的三維建模系統(tǒng),將奠定無人機航測技術規(guī)模化商用的基礎。本文圍繞傾斜影像三維實景建模技術難題,對三維建模技術等關鍵問題進行了系統(tǒng)深入研究,主要包括傾斜式影像空中三角測量技術、多視立體精確三維建模技術、空地融合的二、三維一體化規(guī)劃設計技術,研發(fā)了原型系統(tǒng),開展了無人機航拍影像和車載影像及手持相機影像融合建模試驗。主要工作包括以下幾個方面:1、運用三維數(shù)據(jù)采集和數(shù)據(jù)處理法建立了二三維一體化規(guī)劃模型,揭示了傾斜攝影序列影像多視、多角度、大傾角的特性。通過試驗,分析了傳統(tǒng)影像匹配方法無法解決立體影像匹配系統(tǒng)中的遮擋、幾何變形、幾何斷裂、影像大幅度旋轉等瓶頸問題對土地整治規(guī)劃的影響。確定了適宜開展基于多源測量數(shù)據(jù)融合的三維實景重建技術的條件:(1)海量精確的大場景三維數(shù)據(jù);(2)精確的紋理數(shù)據(jù);(3)控制點數(shù)據(jù)。2、研究了測圖級精度的三維建模技術方法。結合計算機視覺領域中最新的、半全局密集點云匹配算法,通過建立影像與模型的嚴格對應關系,以多視最小二乘矢量匹配支持下的子像素定位精度,形成由2D邊緣特征,經(jīng)過前交3D矢量投影到圖像2D特征驗證的交互式迭代采集模式,同步生成測圖級精度的二、三維矢量成果。3、根據(jù)多源、可量測影像數(shù)據(jù)集,構建基于多視影像的渲染環(huán)境,提供用戶全息瀏覽及三維實景支持下的交互建模體驗,并利用多層次分級建模技術獲取空間對象的精細三維模型:利用航空傾斜影像建立中級模型,采用移動采集影像進行建筑側面及地面景觀精細建模,根據(jù)傾斜攝影和移動測量組合數(shù)據(jù)源,克服空地影像在分辨率、攝影角度、影像輻射特性等方面的差異,構建基于多視影像的攝影測量環(huán)境,提供用戶全息瀏覽及三維實景支持下的交互建模,獲取空間對象的精細三維模型。4、通過三維場景立體、多角度、全方位的展示技術,將土地整治規(guī)劃要素與遙感影像、DEM數(shù)據(jù)、三維實體及實景影像等多源地表數(shù)據(jù)基于統(tǒng)一空間尺度進行融合集成,研究集成化數(shù)據(jù)成果的二、三維一體化系統(tǒng)。5、根據(jù)理論及算法的研究成果實現(xiàn)可規(guī)�；囼灥脑拖到y(tǒng),結合更全面的數(shù)據(jù)資料,不斷吸收國內(nèi)外三維建模處理研究領域取得的最新成果,進一步完善理論模型。本文的創(chuàng)新點簡單概括為以下三個:1、空地聯(lián)合影像定向配準技術利用無人機低空傾斜影像數(shù)據(jù)和地面車載及背負式RTK相機移動采集的地表影像數(shù)據(jù),進行聯(lián)合定向配準,克服傳統(tǒng)空中三角測量算法只能針對下視攝影的困難和傾斜影像存在影像斷裂及遮擋等不連續(xù)問題,并利用圖形處理單元(GPU)并行算法提高大規(guī)模數(shù)據(jù)求解方程組的效率。2、空地融合精細三維實景建模關鍵技術根據(jù)傾斜攝影和移動測量組合數(shù)據(jù)源,克服空地影像在分辨率、攝影角度、影像輻射特性等方面的差異,構建基于多視影像的攝影測量環(huán)境,提供三維實景環(huán)境中的交互式精細三維建模并提供全息瀏覽、查詢。3、三維實景二三維一體化土地整治規(guī)劃設計模式通過空地融合精細三維實景建模技術,構建適合全地形的空地一體化實景模型,提出了空間精確配準的三維實景土地整治規(guī)劃設計模式,徹底改變傳統(tǒng)模式中的人工測量和二維平面規(guī)劃設計方式,提高測量和規(guī)劃設計的效率,增強項目區(qū)規(guī)劃設計效果的直觀展示,強化三維模型在土地整治規(guī)劃設計中的推廣應用。
[Abstract]:The use of tilt photography technology is the current UAV. Some common methods for aerial tilt photography technology can realize multi angle image acquisition, to ensure high efficiency, increase precision, accurate positioning, to achieve a real three-dimensional reconstruction, has become the domestic and foreign non main technical aerial man-machine approach, and gradually applied in the field of geophysical surveying and mapping technology. Is the implementation of land use planning. The basis and key of land consolidation planning at present mainly in the whole field of artificial measurement and traditional two-dimensional graphic design, there are field workload, the mapping efficiency is low, high operation cost, the UAV aerial has become an effective way to solve the above problems, research for 3D modeling system based on tilt photogrammetry no, will lay the scale of aerial man-machine technology commercial foundation. This paper focuses on the modeling technology of 3D virtual image tilt problem, to build three-dimensional Make a systematic research on key issues die technology, including oblique aerial triangulation technique, multi view stereo 3D modeling technology, the technology of 3D space integration two, integrated planning and design, research and development of a prototype system, carry out the UAV aerial image and vehicle images and hand-held camera image fusion. The main modeling test the work includes the following aspects: 1, establish two dimensional integrated planning model using 3D data acquisition and data processing method, reveals the oblique photography image sequence multi view, multi angle, especially the great angle. Through the experiment, occlusion, the traditional image matching method can not solve the stereo image matching system geometric distortion analysis fracture geometry, image rotation and other bottlenecks of land consolidation planning. To determine the suitable to carry out the real three-dimensional multi-source data fusion based on weight measurement Construction conditions: (1) accurate 3D data massive scenes; (2) texture data accurately; (3) the control point data of.2, the research of 3D modeling technology mapping level accuracy. Combined with the latest in computer vision field, semi global dense point cloud matching algorithm, through strict correspondence relationship between image and model, the least squares support vector multi view matching sub-pixel positioning accuracy, formed by the 2D edge, through interactive iterative acquisition mode before 3D vector projection image 2D feature verification, synchronous generation mapping level accuracy two, three-dimensional vector results of.3, according to sources, can be measured the image data set, constructing the multi view image rendering based on the environment, to provide interactive modeling user browsing and real 3D holographic support experience, fine three dimensional model and using the multi-level hierarchical modeling technique to obtain spatial objects: using aerial tilt The image building intermediate model, the construction side and ground landscape fine modeling using mobile image acquisition, according to the tilt photography and mobile measurement data combination source, overcome the image in the space resolution, camera angles, difference image radiation characteristics and other aspects of the construction of photogrammetry environment based on multi view image, providing interactive browsing and real 3D holographic user modeling under the support of.4, obtain the fine 3D model of spatial objects, the 3D scene three-dimensional, multi angle, all-round display technology, land remediation planning elements and remote sensing image, DEM data, and multi-source three-dimensional real image surface spatial scale based on unified data integration, data integration on the two. Three dimensional integrated.5 system, according to the implementation of the prototype scale test results and algorithm theory, combined with the more comprehensive data, not The latest achievements made at home and abroad to absorb broken 3D modeling research, to further improve the theoretical model. The innovation of this paper summarized as the following three: 1, no surface image data of low altitude UAV and ground vehicle and tilt the image data with RTK camera mobile acquisition by joint space image registration technology combined directional, directional registration, to overcome the traditional triangulation algorithm only for visual photography difficulties and tilt image image of fracture and occlusion discontinuities, and the use of the graphics processing unit (GPU) parallel algorithm to improve the large-scale data.2 efficiency equations, space integration key technology of fine 3D modeling based on real tilt photography and mobile measurement data a combination of source, to overcome the image in the space resolution, camera angle, difference image radiation characteristics and other aspects of the construction of multi view image based on photography The measurement environment, providing interactive modeling of 3D fine 3D virtual environments and provide holographic browsing, query.3, 3D virtual two or three dimensional integrated land planning design mode through the fine 3D modeling of real space integration technology, construct space integrated real terrain model, puts forward the space accurate registration of three-dimensional real land planning design mode, completely change the traditional mode of manual measurement and two-dimensional layout planning and design, improve the efficiency of measurement and planning and design, project planning and design exhibition area enhance visual effect, strengthen the popularization and application of 3D model in land consolidation planning and design.

【學位授予單位】：中國地質大學
【學位級別】：博士
【學位授予年份】：2017
【分類號】：P231

【相似文獻】

相關博士學位論文 前8條

1 闞酉潯;基于多源測量數(shù)據(jù)融合的三維實景重建技術研究[D];中國地質大學;2017年

2 李軍;三維封裝電磁干擾的分析與防護設計[D];浙江大學;2017年

3 柴蒙磊;基于單幅圖像的三維發(fā)型建模技術及其應用[D];浙江大學;2017年

4 鐘承勇;三維碳網(wǎng)絡中的電子性質及拓撲效應[D];湘潭大學;2017年

5 張季一;基于幾何代數(shù)的三維地籍空間數(shù)據(jù)模型研究[D];中國礦業(yè)大學;2016年

6 初文華;處理非連續(xù)問題的三維SPH算法及其在沖擊動力學問題中的應用[D];哈爾濱工程大學;2013年

7 劉邈;高精度光學三維標示關鍵技術及應用研究[D];天津大學;2016年

8 高俊義;稀疏不規(guī)則裂隙巖體水流—傳熱對溫度和應力影響的三維模型試驗與離散元模擬研究[D];北京交通大學;2017年

相關碩士學位論文 前8條

1 吳豪杰;清oi河水質三維仿真系統(tǒng)設計與實現(xiàn)[D];鄭州大學;2017年

2 孫營偉;基于高分2號遙感數(shù)據(jù)的三維綠量估算模型研究[D];中國地質大學(北京);2017年

3 陳驊;三維古建筑場景仿真研究[D];西安科技大學;2017年

4 李旭光;基于免疫反應調控細胞在生物材料中的三維遷移行為[D];浙江大學;2017年

5 趙之星;用于三維建模的無人機傾斜攝影飛行參數(shù)優(yōu)化研究[D];西安科技大學;2017年

6 顏麗穎;波導Helix陣三維Fourier變換匹配場定位[D];浙江大學;2017年

7 李娜;無線傳感器網(wǎng)絡三維DV-Hop定位算法研究[D];江西理工大學;2017年

8 楊瑞琳;三維晶體相場法模擬應力作用下晶體的預熔及熔解[D];廣西大學;2017年

，

本文編號：1771135