本研究通過手機拍攝將多張二維圖像處理,轉換為點雲數據後,與 LIDAR
數據結合,生成了一個完整的點雲模型。相較於傳統的高精度三維掃描技術,
我們的方法具有低成本和易實現性的優勢。無需昂貴的設備和複雜的操作,只
需要一部智能手機和 Python 即可實現物體的點雲模型生成。
通過手機拍攝二維圖像轉換為點雲數據,我們實現了對物體的形狀和結構
的快速捕捉。將二維圖像結合 LIDAR 數據,進一步豐富了點雲模型的細節和資
訊。LIDAR 數據提供了更精確的深度信息,使得點雲模型的幾何形狀更加精確
和真實。這種結合有助於在模型生成的過程中克服二維圖像的限制,提高了建
模的準確性和可靠性。
生成出的點雲模型應用:點雲模型可以用於可視化、分析、測量和模型修
復等多個方面。最後生成出的模型和現實的物體只有微小的差距,如果不是應
用在要求高精度的模型,這樣生成的模型是可用的。

In this paper, utilizes mobile phone photography to capture multiple two-
dimensional images, which are processed and transformed into point cloud data. By combining this data with LIDAR data, a complete point cloud model is generated.Compared to traditional high-precision 3D scanning techniques, our method offers advantages in terms of low cost and ease of implementation. It does not require expensive equipment or complex operations; only a smartphone and Python are needed to generate the point cloud model of an object.
Through the conversion of two-dimensional images captured by a mobile phone
into point cloud data, we achieve rapid capture of the object's shape and structure. The integration of two-dimensional images with LIDAR data further enriches the details and information of the point cloud model. LIDAR data provides more accurate depth information, resulting in a more precise and realistic geometric shape of the point cloud model. This combination helps overcome the limitations of two-dimensional images in the modeling process, improving accuracy and reliability.
Applications of the generated point cloud model include visualization, analysis,
measurement, and model repair, among other aspects. The resulting model exhibits
only minor differences compared to the real object, making it usable unless high-precision models are required.

摘要 i
Abstract ii
致謝 iv
目錄 v
表目錄 vii
圖目錄 viii
第一章　緒論 1
1.1前言 1
1.2文獻回顧 2
1.3研究動機與目的 5
第二章　研究方法與理論 6
2.1二維RGB圖像原理 6
2.2深度圖原理 8
2.3 LIDAR光學雷達原理 10
2.4 相機座標系 11
2.4.1 二維轉三維座標原理12
2.5 Python程式語言 14
2.5.1 NumPy 15
2.5.2 OpenCV 16
2.5.3 Open3D 17
2.6 點雲圖 18
2.7 表面重建 18
2.8 Mesh模型 19
第三章　實驗內容 20
3.1分析二維圖 21
3.1.1 獲取二維圖像 21
3.1.2 二維圖像高斯模糊處理 22
3.1.3 灰階化處理 25
3.1.4 輪廓圖處理 28
3.2深度圖提取 31
3.2.1提取深度圖 31
3.2.2 提取深度圖像素座標 33
3.3 點雲圖生成和處理 35
3.3.1 生成點雲 35
3.3.2 創建點雲模型 38
3.3.3 點雲模型背景處理 40
3.4 LIDAR結合點雲模型 42
3.5 生成Mesh模型 43
3.4 表面重建 45
第四章　實驗結果 48
4.1 Mesh模型比較 48
4.2 表面還原比較 49
4.3 模型尺寸 50
第五章　結論與未來展望 52
5.1 結論 52
5.2 未來展望 53
參考文獻 54

[1] H. Barrow and J. Tenenbaum. Recovering intrinsic scene characteristics.
Comput.Vis. Syst., A Hanson & E. Riseman (Eds.), pages 3–26, 1978.
[2] A. Saxena, S. H. Chung, and A. Ng. Learning depth from single monocular
images. In Neural Information Processing Systems (NIPS), volume 18, pages
1161, 2005.
[3] H. Su, C. R. Qi, Y. Li, and L. J. Guibas. Render for cnn: Viewpoint estimation
in images using cnns trained with rendered 3d model views. In Proceedings of
the IEEE International Conference on Computer Vision, pages 2686–2694,
2015.
[4] J. T. Barron and J. Malik. Shape, illumination, and reflectance from shading.
TPAMI, 2015.
[5] Lequan Yu, Xianzhi Li, Chi-Wing Fu, Daniel Cohen-Or, Pheng-Ann Heng.
PU-Net: Point Cloud Upsampling Network, 2018.
[6] Sharma, A., & Choudhury, T. Python-based 3D shape reconstruction from 2D
images. In Proceedings of the International Conference on Advances in
Computing, Communication and Control, pages 198-204, 2018.
[7] Gonzalez, R. C., & Woods, R. E. Digital Image Processing (4th ed.). Pearson,
2018.
[8] Fergus, R., Singh, B., Hertzmann, A., Roweis, S. T., & Freeman, W. T.
Removing camera shake from a single photograph. ACM Transactions on
Graphics (TOG), 26(3), pages 1-8, 2007.
[9] A. Ismail and A. Okasha, "Lidar Remote Sensing: An Introduction and
Overview," Journal of Environmental Science and Technology, vol. 8, no. 1,
pages 29-52, 2015.
[10] Szeliski, R. Computer Vision: Algorithms and Applications. Springer, 2010.
(計算機視覺:算法與應用)
[11] Zhang, Z. A flexible new technique for camera calibration.
IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(11),
pages 1330-1334, 2000.
[12] VanderPlas, J. Python Data Science Handbook: Essential Tools for. Working
with Data. O'Reilly Media, 2016.
[13] Dutta, R., & Mandal, S. Application of NumPy in data mining. International
Journal of Computer Science and Network Security, 18(2), pages 139-146,
2018.
[14] Bradski, G., & Kaehler, A. Learning OpenCV: Computer Vision with the
OpenCV Library. O'Reilly Media, 2008.
[15] Qian-Yi Zhou, Jaesik Park, and Vladlen Koltun. "Open3D: A Modern Library
for 3D Data Processing." arXiv preprint arXiv:1801.09847, 2018.
[16] Rusu, R. B., & Cousins, S. 3D is here: Point Cloud Library (PCL). In IEEE
International Conference on Robotics and Automation (ICRA), pages 1-4,
2011.
[17] Berger, M. O., & Bokeloh, M. Surface reconstruction from point clouds. In
Eurographics 2016-Tutorials, 2016.
[18] Tomasi, C., & Manduchi, R. Bilateral filtering for gray and color. images. In
IEEE International Conference on Computer Vision, pages 839-846, 1998.
[19] Alexa, M., Behr, J., Cohen-Or, D., Fleishman, S., Levin, D., & Silva, C. T.
Computing and rendering point set surfaces. IEEE Transactions on
Visualization and Computer Graphics, 7(4), pages 367-375, 2001.