在本篇論文中，介紹一個利用深度學習將三維模型進行對稱分割，並且產生線條藝術的方法。本作品的起源是在於藝術家製作線條藝術時，會根據實體的模型外圍來進行線條生成與組合，最後進而與原本模型實體相當。本篇整體概要是在於如何用同樣的方式基於深度學習的模型分割對3D模型來產生線條藝術。首先，我們利用深度學習的方法來對模型的網格(mesh)先進行分割，也就是先對模型本身來進行分類，之後才能達到對稱線條的結果。接著萃取網格上的邊來進行合併(merge)，並產生合併後的邊線段，這些線段大致上就可以來表示模型的外圍輪廓。產生線段後，為了使連接後的連續線條達到最少的數量，我們制訂了多旅行商問題(multiple traveling salesman problem, mTSP)來達成目的進而產生線條藝術。我們發現線條生成會產生不連續線段的斷成，因此也制定了根據使用者選取不同線段之間的節點(node)來進行合併，此方式可以讓最後的成品看起來較自然。我們希望最後的線條能夠盡量達到平滑的效果，對此我們利用一個最佳化的方式使線段之間角度差別較大的地方進行平滑化，使得最後生成的線條不會看起來崎嶇不平，而較有連續感。另外、我們也提供了介面可以依照個人喜好來進行線段連接，產生符合使用者的預期結果。

In this paper, we introduce a method of using deep learning to symmetrically segment a three-dimensional model and generate line art. The origin of this work is that when the artist makes line art, it will generate and combine lines based on the periphery of the entity's model, and finally it will be equivalent to the original model entity. The overall summary of this article is how to use the same way to generate line art based on the model segmentation of deep learning on the 3D model. First, we use the deep learning method to segment the mesh of the model first, that is, to classify the model itself before reaching the result of symmetrical lines. Then extract the edges on the mesh to merge, and generate merged edge line segments, which can roughly represent the outer contour of the model. After generating line segments, in order to minimize the number of connected continuous lines, we formulated the multiple traveling salesman problem (mTSP) to achieve the goal and produce line art. We found that line generation will produce discontinuous line segments. Therefore, we have also formulated a combination of nodes between different line segments selected by the user. This method can make the results look more natural. We hope that the results of line art can be as smooth as possible. For this, we use an optimized method to smooth the parts where the angle difference between the line segments is large, so that the final generated lines will not look rugged, but with smooth lines. In addition, we also provide an interface to connect line segments according to personal preferences, and produce results that meet the user's expectations.

[1] Hsiao, Kai-Wen, Jia-Bin Huang, and Hung-Kuo Chu. "Multi-view wire art." ACM Trans. Graph. 2018.
[2] Wang, P., Gan, Y., Shui, P., Yu, F., Zhang, Y., Chen, S., & Sun, Z. 3D shape segmentation via shape fully convolutional networks. Computers & Graphics, 76, 182-192. 2018
[3] Lingjie Liu, Duygu Ceylan, Cheng Lin, Wenping Wang, and Niloy J. Mitra. Image-based reconstruction of wire art. ACM Trans. Graph., 36(4):63:1–63:11, July 2017.
[4] YEH, CHIH-KUO, et al. "Generating Virtual Wire Sculptural Art from 3D Models." 2021.
[5] Andrew Nealen, Takeo Igarashi, Olga Sorkine, and Marc Alexa. Fibermesh: Designing freeform surfaces with 3d curves. In ACM SIGGRAPH 2007 Papers, SIGGRAPH ’07, New York, NY, USA, 2007. ACM.
[6] Hanocka, R., Hertz, A., Fish, N., Giryes, R., Fleishman, S., & Cohen-Or, D. Meshcnn: a network with an edge. ACM Transactions on Graphics (TOG), 38(4), 1-12. 2019.
[7] Michael Garland and Paul S. Heckbert. Surface simplification using quadric error metrics. In Proceedings of the 24th Annual Conference on Computer Graphics and Interactive Techniques, SIGGRAPH ’97, pages 209–216, New York, NY, USA, 1997. ACM Press/Addison-Wesley Publishing Co.
[8] Emil Ernerfeldt. Fitting a plane to noisy points in 3d, September 2017.
[9] Imdat Kara and Tolga Bektas. Integer linear programming formulations of multiple salesman problems and its variations. European Journal of Operational Research, 174(3):1449 – 1458, 2006.
[10] Rusinkiewicz, S. Estimating curvatures and their derivatives on triangle meshes. In Proceedings. 2nd International Symposium on 3D Data Processing, Visualization and Transmission, 2004. 3DPVT 2004. (pp. 486-493). IEEE. 2004, September.
[11] Lira, W., Fu, C. W., & Zhang, H. Fabricable eulerian wires for 3D shape abstraction. ACM Transactions on Graphics (TOG), 37(6), 1-13. 2018.
[12] Huang, Zhihu, and Jinsong Leng. "Analysis of Hu's moment invariants on image scaling and rotation." 2010 2nd International Conference on Computer Engineering and Technology. Vol. 7. IEEE, 2010.