随着倾斜摄影三维重建技术的发展，城市场景三维模型的应用需求及空间范围都在不断增加，同时高度细节化的模型表达和高分辨率配套纹理更使模型数据量日趋庞大，给计算机渲染、存储及传输都带来了巨大挑战。为解决此问题，本文采用LOD(层次细节模型，Level of Detail)技术方案，能够均衡模型精度与计算机渲染速度，为用户带来更为流畅的浏览体验。因此本文主要研究内容如下：

（1）针对城市场景三维模型因数据量庞大常使用瓦片分割技术所遗留的原始模型色彩不连续问题，采用基于泊松编辑改进的瓦片间匀光算法。相较传统方法，该算法以三角面为基本单位，能够直击模型表面，有效消除瓦片间色差。

（2）针对传统简化算法重几何轻纹理特征从而导致用于构建LOD模型的层级简化网纹理质量不佳问题，提出完整高效的简化网纹理重建算法。通过借助原始网分块信息实现简化网快速分割后，以分割纹理单元为单位进行参数化，并将所得二维纹理块组织为纹理图集，最后直接取原始网像素信息并赋予简化网。因大量借助原始网信息，算法效率及简化网纹理质量均有显著提升。针对简化网配套纹理图集总体图幅较大问题，提出基于CUDA多线程并行运算策略的纹理排列算法，混合使用矩形排列和多边形排列方式，并建立多套模板进行加速，有效提升了纹理图集空间利用率及算法效率。

（3）针对因文件数据量大而导致LOD模型浏览不流畅问题，采用基于数据量自适应和局部切片简化比率自适应的节点优化策略。通过深度优先遍历方式获取当前节点以及兄弟节点的总文件数据量，并计算与父节点切片模型的简化比率，若达到阈值则合并兄弟节点并以此替换父节点，从而达到删除冗余数据的目的。同时，得益于本文高密度纹理排列算法加持，更使模型轻量化。最后经动态调度，实现模型流畅显示。

以倾斜摄影影像构建的三维Mesh模型作为实验数据，与现有方法、开源软件（OpenMVE、XAtlas）及商业软件（Context Capture）进行对比，结果表明：采用本文方法进行简化网纹理重建不但纹理质量高且计算效率也有显著提升。纹理排列复合空间利用率达99.766%。并且LOD模型平均浏览帧率高、模型浏览更加流畅。

With the development of oblique photography 3D reconstruction technology, the application requirements and spatial scope of 3D models of urban scenes are constantly increasing. At the same time, the highly detailed model expression and high-resolution supporting textures make the model data increasingly huge, which is used for computer rendering, Both storage and transmission present huge challenges. In order to solve this problem, this paper adopts LOD (level of detail) technology scheme, which can balance the model accuracy and computer rendering speed, and bring users a more smooth browsing experience. Therefore, the main research contents of this paper are as follows:

(1) Aiming at the color discontinuity of the original model left by the tile segmentation technology often used in the 3D model of urban scene due to the large amount of data, an improved dodging method based on Poisson editing is proposed. Compared with the traditional method, the algorithm takes the triangular surface as the basic unit, which can directly hit the model surface and effectively eliminate the color difference between tiles.

(2) Aiming at the problem that the traditional simplification algorithm emphasizes geometry and light texture features, which leads to the poor quality of the hierarchical simplified mesh used to construct the LOD model, a complete and efficient simplified mesh texture reconstruction algorithm is proposed. After the fast segmentation of the simplified mesh is realized by using the texture connected region information of the original mesh, the segmented texture unit is parameterized, and the obtained 2D texture blocks are organized into a texture atlas. Finally, the pixel information is obtained directly from the original mesh and given to the simplified mesh. With the help of a large number of original mesh information, the efficiency of the algorithm and the quality of simplified mesh texture are significantly improved. Aiming at the problem that the overall map of the texture atlas supporting the simplified mesh is large, a texture packing algorithm based on CUDA multi-threaded parallel operation strategy is proposed, which uses a mixture of rectangular packing and polygon packing, and establish multiple sets of templates for acceleration, which effectively improves the space utilization and algorithm efficiency of the texture atlas.

(3) Aiming at the problem that the browsing of LOD model is not smooth due to the large amount of file data, a node optimization strategy based on adaptive data volume and adaptive local slice simplification ratio is proposed. Obtain the total file data of the current node and brother nodes by depth first search, and calculate the simplified ratio with the slice model of the parent node. If the threshold is reached, merge the brother nodes and replace the parent node, so as to delete redundant data. Moreover, due to the high-density texture packing algorithm, the model is lighter. Finally, through dynamic scheduling, the smooth display of the model is realized.

Taking the 3D mesh model constructed from oblique photographic images as the experimental data, it is compared with the existing methods, open source software (OpenMVE、XAtlas) and commercial software (Context Capture). The results show that the simplified mesh texture reconstruction using this method not only has high texture quality, but also significantly improves the calculation efficiency. The utilization rate of texture packing composite space is 99.766%. And the average browsing frame rate of LOD model is higher, so the browsing of model is more smooth.

[1] Clark J H. Hierarchical geometric models for visible surface algorithms[J]. Communications of the ACM, 1976, 19(10): 547-554.

[2] Rubin S M. The representation and display of scenes with a wide range of detail[J]. Computer Graphics and Image Processing, 1982, 19(3): 291-298.

[3] Eck M, DeRose T, Duchamp T, et al. Multiresolution analysis of arbitrary meshes[C]//Proceedings of the 22nd annual conference on Computer graphics and interactive techniques. 1995: 173-182.

[4] Lindstrom P, Koller D, Ribarsky W, et al. Real-time, continuous level of detail rendering of height fields[C]//Proceedings of the 23rd annual conference on Computer graphics and interactive techniques. 1996: 109-118.

[5] Lindstrom P, Pascucci V. Visualization of large terrains made easy[C]//Proceedings Visualization, 2001. VIS'01. IEEE, 2001: 363-574.

[6] Lindstrom P, Pascucci V. Terrain simplification simplified: A general framework for view-dependent out-of-core visualization[J]. IEEE Transactions on Visualization and Computer graphics, 2002, 8(3): 239-254.

[7] Duchaineau M, Wolinsky M, Sigeti D E, et al. ROAMing terrain: Real-time optimally adapting meshes[C]//Proceedings. Visualization'97 (Cat. No. 97CB36155). IEEE, 1997: 81-88.

[8] Blow J. Terrain rendering at high levels of detail[C]//Proceedings of the 2000 Game Developers Conference. sn, 2000, 3.

[9] Hoppe H. View-dependent refinement of progressive meshes[C]//Proceedings of the 24th annual conference on Computer graphics and interactive techniques. 1997: 189-198.

[10] Ji J, Wu E, Li S, et al. View-dependent refinement of multiresolution meshes using programmable graphics hardware[J]. The Visual Computer, 2006, 22(6): 424-433.

[11] DeCoro C, Tatarchuk N. Real-time mesh simplification using the GPU[C]//Proceedings of the 2007 symposium on Interactive 3D graphics and games. 2007: 161-166.

[12] Peng C, Cao Y. A GPU‐based approach for massive model rendering with frame‐to‐frame coherence[C]//Computer Graphics Forum. Oxford, UK: Blackwell Publishing Ltd, 2012, 31(2pt2): 393-402.

[13] Lodha S K, Roskin K M, Renteria J C. Hierarchical topology-preserving simplification of terrains[J]. The Visual Computer, 2003, 19(7): 493-504.

[14] Wu J, Yang Y, Gong S, et al. A New Quadtree-based Terrain LOD Algorithm[J]. J. Softw., 2010, 5(7): 769-776.

[15] Kang L, Xu J, Yang C, et al. An efficient simplification and real-time rendering algorithm for large-scale terrain[J]. International journal of computer applications in technology, 2010, 38(1-3): 106-112.

[16] Lee E S, Lee J H, Shin B S. Bimodal vertex splitting: acceleration of quadtree triangulation for terrain rendering[J]. IEICE TRANSACTIONS on Information and Systems, 2014, 97(6): 1624-1633.

[17] Kalem S A, Kourgli A. Large-scale terrain level of detail estimation based on wavelet transform[C]//Proceedings of the 10th International Conference on Computer Graphics Theory and Applications. 2015: 258-264.

[18] Trapp M, Döllner J. Interactive close-up rendering for detail+ overview visualization of 3D digital terrain models[C]//2019 23rd International Conference Information Visualisation (IV). IEEE, 2019: 275-280.

[19] 赵友兵, 石教英, 周骥, 等. 一种大规模地形的快速漫游算法[J]. 计算机辅助设计与图形学学报, 2002, 14(7): 624-628.

[20] 余明, 左小清, 李清泉. 一种基于TIN的视相关动态多分辨率地形模型[J]. 武汉大学学报(信息科学版), 2004(12): 1106-1110.

[21] 谭兵, 徐青, 马东洋. 用约束四叉树实现地形的实时多分辨率绘制[J]. 计算机辅助设计与图形学学报, 2003(3): 270-276.

[22] 张继开, 古梅, 黄心渊. 三维地形可视化技术的研究[J]. 计算机仿真, 2005(07): 118-119+148.

[23] 张桀宁, 李帅. 一种基于顶点纹理的LOD地形渲染算法[J]. 系统仿真学报, 2008(07): 1758-1764.

[24] 彭海. 三维地形可视化技术的研究与实现[D].武汉理工大学,2010.

[25] 李钦, 戴树岭, 赵永嘉, 等. 分块 LOD 大规模地形实时渲染算法[J]. 计算机辅助设计与图形学学报, 2013, 25(5): 708-713.

[26] 王青云, 罗泽. 基于LOD的海量地形数据并行渲染技术[J]. 计算机系统应用, 2017, 26(12): 200-206.

[27] 陈兴旺. 视觉感知驱动的复杂三维城市场景数据自适应组织与调度方法[D]. 西南交通大学, 2018.

[28] Cao W, Huang L, Hu Y, et al. An Improved Algorithm for Terrain Rendering[C]//IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2019: 3452-3455.

[29] 李烁, 王慧, 王利勇, 等. 遥感影像变分Mask自适应匀光算法[J]. 遥感学报, 2018, 22(3): 450-457.

[30] 刘小平, 赖剑煌, 张智斌. 基于小波子带图像的人脸光照归一化方法[J]. 中山大学学报(自然科学版), 2007(5): 25-28.

[31] 孙文, 尤红建, 傅兴玉, 等. 基于非线性MASK的遥感影像匀光算法[J]. 测绘科学, 2014, 39(9): 130-134.

[32] 张力, 张祖勋, 张剑清. Wallis滤波在影像匹配中的应用[J]. 武汉测绘科技大学学报, 1999(1): 24-27+35.

[33] Reinhard E, Adhikhmin M, Gooch B, et al. Color transfer between images[J]. IEEE Computer graphics and applications, 2001, 21(5): 34-41.

[34] Xiao X, Ma L. Color transfer in correlated color space[C]//Proceedings of the 2006 ACM international conference on Virtual reality continuum and its applications. 2006: 305-309.

[35] Ilie A, Welch G. Ensuring color consistency across multiple cameras[C]//Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1. IEEE, 2005, 2: 1268-1275.

[36] Yoo J D, Park M K, Cho J H, et al. Local color transfer between images using dominant colors[J]. Journal of Electronic Imaging, 2013, 22(3): 033003.

[37] Nguyen R M H, Kim S J, Brown M S. Illuminant aware gamut‐based color transfer[C]//Computer Graphics Forum. 2014, 33(7): 319-328.

[38] Frigo O, Sabater N, Demoulin V, et al. Optimal transportation for example-guided color transfer[C]//Asian Conference on Computer Vision. Springer, Cham, 2014: 655-670.

[39] Wang B, Yu Y, Xu Y Q. Example-based image color and tone style enhancement[J]. ACM Transactions on Graphics (TOG), 2011, 30(4): 1-12.

[40] Brown M, Lowe D G. Automatic panoramic image stitching using invariant features[J]. International journal of computer vision, 2007, 74(1): 59-73.

[41] Xiong Y, Pulli K. Color matching of image sequences with combined gamma and linear corrections[C]//International Conference on ACM Multimedia. 2010: 261-270.

[42] Shen T, Wang J, Fang T, et al. Color correction for image-based modeling in the large[C]//Asian Conference on Computer Vision. Springer, Cham, 2016: 392-407.

[43] Schroeder W J, Zarge J A, Lorensen W E. Decimation of triangle meshes[C]//Proceedings of the 19th annual conference on Computer graphics and interactive techniques. 1992: 65-70.

[44] Cohen J, Varshney A, Manocha D, et al. Simplification envelopes[C]//Proceedings of the 23rd annual conference on Computer graphics and interactive techniques. 1996: 119-128.

[45] Hoppe H, DeRose T, Duchamp T, et al. Mesh optimization[C]//Proceedings of the 20th annual conference on Computer graphics and interactive techniques. 1993: 19-26.

[46] Hoppe H. Progressive meshes[C]//Proceedings of the 23rd annual conference on Computer graphics and interactive techniques. 1996: 99-108.

[47] Garland M, Heckbert P S. Surface simplification using quadric error metrics[C]//Proceedings of the 24th annual conference on Computer graphics and interactive techniques. 1997: 209-216.

[48] 刘晓利, 刘则毅, 高鹏东, 等. 基于尖特征度的边折叠简化算法[J]. 软件学报, 2005(5): 669-675.

[49] 刘焕敏, 杨克俭, 王玉华. 一种面积加权的半边折叠网格简化算法及其递进网格构造[J]. 武汉理工大学学报(交通科学与工程版), 2005(1): 76-78+82.

[50] Ozaki H, Kyota F, Kanai T. Out-of-Core Framework for QEM-based Mesh Simplification[C]//EGPGV@ EuroVis. 2015: 87-96.

[51] 李胜辉, 史瑞芝, 朱明. 一种顾及多重约束的三维地形简化方法[J]. 武汉大学学报(信息科学版), 2018, 43(2): 241-247.

[52] Hamann B. A data reduction scheme for triangulated surfaces[J]. Computer aided geometric design, 1994, 11(2): 197-214.

[53] Garland M, Heckbert P S. Simplifying surfaces with color and texture using quadric error metrics[C]//Proceedings Visualization'98 (Cat. No. 98CB36276). IEEE, 1998: 263-269.

[54] Cohen J, Manocha D, Olano M. Simplifying polygonal models using successive mappings[C]//Proceedings. Visualization'97 (Cat. No. 97CB36155). IEEE, 1997: 395-402.

[55] 冯翔, 周明全. 带纹理的三维模型简化算法[J]. 计算机辅助设计与图形学学报, 2009, 21(6): 842-846+852.

[56] Sporysz P, Głomb P. Improving the quality of mesh simplification with texture saliency measurement[C]//2014 4th International Conference on Image Processing Theory, Tools and Applications (IPTA). IEEE, 2014: 1-6.

[57] 李世俊, 姜晓彤, 唐慧. 保持细节特征的带纹理模型的高质量简化算法[J]. 计算机应用研究, 2020, 37(1): 300-303+312.

[58] Cignoni P, Montani C, Rocchini C, et al. Preserving attribute values on simplified meshes by resampling detail textures[J]. The Visual Computer, 1999, 15(10): 519-539.

[59] Cheng I, Boulanger P. Scale-space 3D TexMesh simplification[C]//2004 IEEE International Conference on Multimedia and Expo (ICME)(IEEE Cat. No. 04TH8763). IEEE, 2004, 1: 141-144.

[60] Garcia I, Patow G. Igt: inverse geometric textures[M]//ACM SIGGRAPH Asia 2008 papers. 2008: 1-9.

[61] 黄翔翔, 朱全生, 江万寿. 多视纹理映射中无需设定偏差的快速可见性检测[J]. 测绘学报, 2020, 49(1): 92-107.

[62] 叶春阳. 网格参数化的高效优化方法研究[D]. 中国科学技术大学, 2021.

[63] Jiang Z, Schaefer S, Panozzo D. Simplicial complex augmentation framework for bijective maps[J]. ACM Transactions on Graphics, 2017, 36(6).

[64] Wang C, Hu X, Fu X, et al. Bijective spherical parametrization with low distortion[J]. Computers & Graphics, 2016, 58: 161-171.

[65] Daniels J, Silva C T, Cohen E. Semi‐regular quadrilateral‐only remeshing from simplified base domains[C]//Computer Graphics Forum. Oxford, UK: Blackwell Publishing Ltd, 2009, 28(5): 1427-1435.

[66] Zhu T, Ye C, Chai S, et al. Greedy cut construction for parameterizations[C]//Computer Graphics Forum. 2020, 39(2): 191-202.

[67] Tutte W T. How to draw a graph[J]. Proceedings of the London Mathematical Society, 1963, 3(1): 743-767.

[68] Eck M, DeRose T, Duchamp T, et al. Multiresolution analysis of arbitrary meshes[C]//Proceedings of the 22nd annual conference on Computer graphics and interactive techniques. 1995: 173-182.

[69] Lévy B, Petitjean S, Ray N, et al. Least squares conformal maps for automatic texture atlas generation[J]. ACM transactions on graphics (TOG), 2002, 21(3): 362-371.

[70] Liu L, Zhang L, Xu Y, et al. A local/global approach to mesh parameterization[C]//Computer Graphics Forum. Oxford, UK: Blackwell Publishing Ltd, 2008, 27(5): 1495-1504.

[71] Bright A, Chien E, Weber O. Harmonic global parametrization with rational holonomy[J]. ACM Transactions on Graphics (TOG), 2017, 36(4): 1-15.

[72] Milenkovic V J. Rotational polygon containment and minimum enclosure using only robust 2D constructions[J]. Computational Geometry, 1999, 13(1): 3-19.

[73] Imamichi T, Yagiura M, Nagamochi H. An iterated local search algorithm based on nonlinear programming for the irregular strip packing problem[J]. Discrete Optimization, 2009, 6(4): 345-361.

[74] Leung S C H, Lin Y, Zhang D. Extended local search algorithm based on nonlinear programming for two-dimensional irregular strip packing problem[J]. Computers & Operations Research, 2012, 39(3): 678-686.

[75] Elkeran A. A new approach for sheet nesting problem using guided cuckoo search and pairwise clustering[J]. European Journal of Operational Research, 2013, 231(3): 757-769.

[76] Martinez-Sykora A, Alvarez-Valdés R, Bennell J A, et al. Matheuristics for the irregular bin packing problem with free rotations[J]. European Journal of Operational Research, 2017, 258(2): 440-455.

[77] Abeysooriya R P, Bennell J A, Martinez-Sykora A. Jostle heuristics for the 2D-irregular shapes bin packing problems with free rotation[J]. International Journal of Production Economics, 2018, 195: 12-2