學術產出-Theses
Article View/Open
Publication Export
-
題名 以圖神經網路將 2.5D 樂高建構映射至平鋪問題之方法
Mapping 2.5D Lego Construction into Tiling Problem with Graph Neural Network作者 黃威
Huang, Wei貢獻者 紀明德
Chi, Ming-Te
黃威
Huang, Wei關鍵詞 樂高
圖神經網路
超像素問題
LEGO
Graphic neural network,
Superpixel日期 2024 上傳時間 1-Mar-2024 13:42:18 (UTC+8) 摘要 樂高公司以積木的多樣性深受大人和小孩喜愛，隨著模型複雜度 的增加，人們對樂高模型的組裝有了更高的要求。以樂高浮雕系列為 例，模型以其高度立體的設計和細緻的細節而聞名，處理複雜的三維 空間和結構問題上的能力，使組裝過程更具挑戰性。 本研究著重在處理樂高浮雕系列的複雜性。在這一過程中，我們 需要克服積木的幾何形狀、分層架構和結構強度等多重挑戰，同時必 須在有限的樂高磚資源下實現豐富多樣的創意。為了解決這些問題， 我們採用了三項關鍵技術：圖像分層、樂高生成技術與相似度量化分 析。首先，透過圖像分層技術，我們得以細緻地將輸入圖像分為前景 和背景，深入切分圖像中的細節，進而突顯更多層次的圖像細節。其 次，我們應用樂高生成技術，在區域內最大化平鋪樂高磚，確保模型 的結構穩固，同時解決超像素問題。最後，我們運用相似度量化分析 演算法來比較生成的模型和原始輸入圖像的相似度，全面評估和比較 各種模型的表現。這項分析不僅確保了模型的忠實還原，同時也為我 們提供了改進的空間，以進一步提高模型的精確度和真實感。這些技 術的綜合應用為樂高浮雕系列的設計提供了全新的方法和解決方案， 進一步滿足了樂高愛好者的多樣性。
The LEGO company’s diverse building blocks are loved by both adults and children. As models become more complex, there are higher demands for assembling LEGO models. For example, the LEGO relief series, known for its intricate three-dimensional design , presents challenges in handling complex spatial and structural issues. Our study focuses on addressing the complexity of the LEGO relief series. We employ three key technologies: image segmentation, LEGO generation techniques, and similarity quantification analysis. Image segmentation divides input images into foreground and background, emphasizing more layers of detail. LEGO generation techniques maximize brick placement for structural stability while solving the superpixel problem. Similarity quantification analysis ensures faithful reproduction of models and provides room for improvement.By applying these technologies, we offer new methods and solutions for designing LEGO relief series, catering to the diverse interests of LEGO enthusiasts.參考文獻 [1] LEGO® Starry Night. https://www.lego.com/zh-tw/categories/adults-welcome/ article/details-of-van-gogh-starry-night. [2] LEGO® Great Wave. https://www.lego.com/zh-tw/categories/adults-welcome/ article/how-we-made-the-lego-great-wave. [3] LEGO® Wind God and Thunder God Screens. https://toymim.com/review/ lego-store-narita-airport-report-2020-01. [4] A. Rivers, T. Igarashi, and F. Durand, “2.5 d cartoon models,” ACM Transactions on Graphics (TOG), vol. 29, no. 4, pp. 1–7, 2010. [5] H. Xu, K. H. Hui, C.-W. Fu, and H. Zhang, “Tilingnn: learning to tile with selfsupervised graph neural network,” arXiv preprint arXiv:2007.02278, 2020. [6] LEGO® Brick Modified . https://rebrickable.com/parts/87087/ brick-special-1-x-1-with-stud-on-1-side/. [7] R. Ranftl, K. Lasinger, D. Hafner, K. Schindler, and V. Koltun, “Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer,” IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2020. 43 [8] R. Ranftl, A. Bochkovskiy, and V. Koltun, “Vision transformers for dense prediction,” ArXiv preprint, 2021. [9] Z. Wu, S. Pan, F. Chen, G. Long, C. Zhang, and S. Y. Philip, “A comprehensive survey on graph neural networks,” IEEE transactions on neural networks and learning systems, vol. 32, no. 1, pp. 4–24, 2020. [10] L. Sacht, “Structure-aware bottle cap art,” Computers & Graphics, vol. 107, pp. 277–288, 2022. [11] J. Allebach and P. W. Wong, “Edge-directed interpolation,” in Proceedings of 3rd IEEE International Conference on Image Processing, vol. 3. IEEE, 1996, pp. 707–710. [12] R. E. Carlson and F. N. Fritsch, “Monotone piecewise bicubic interpolation,” SIAM journal on numerical analysis, vol. 22, no. 2, pp. 386–400, 1985. [13] 翁瑋辰, “具樂高平滑化之影像樂高風格化技術,” 2019. [14] K. He, G. Gkioxari, P. Dollár, and R. Girshick, “Mask r-cnn,” in Proceedings of the IEEE international conference on computer vision, 2017, pp. 2961–2969. [15] 王祥宇, “以圖神經網路將二維樂高建構映射至平鋪問題之方法,” 2022. [16] P. Lei, S. Xu, and S. Zhang, “An art-oriented pixelation method for cartoon images,” The Visual Computer, pp. 1–13, 2023. [17] R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 586–595. 44 [18] R. Gower, A. Heydtmann, and H. Petersen, “Lego: Automated model construction,” 1998. [19] M.-H. Kuo, Y.-E. Lin, H.-K. Chu, R.-R. Lee, and Y.-L. Yang, “Pixel2brick: Constructing brick sculptures from pixel art,” in Computer Graphics Forum, vol. 34, no. 7. Wiley Online Library, 2015, pp. 339–348. [20] S.-J. Luo, Y. Yue, C.-K. Huang, Y.-H. Chung, S. Imai, T. Nishita, and B.-Y. Chen, “Legolization: Optimizing lego designs,” ACM Transactions on Graphics (TOG), vol. 34, no. 6, pp. 1–12, 2015. [21] H. Xu, K.-H. Hui, C.-W. Fu, and H. Zhang, “Computational lego technic design,” arXiv preprint arXiv:2007.02245, 2020. [22] K. Lennon, K. Fransen, A. O’Brien, Y. Cao, M. Beveridge, Y. Arefeen, N. Singh, and I. Drori, “Image2lego: customized lego set generation from images,” arXiv preprint arXiv:2108.08477, 2021. [23] Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, “Image quality assessment: from error visibility to structural similarity,” IEEE transactions on image processing, vol. 13, no. 4, pp. 600–612, 2004. [24] M.-R. Huang and R.-R. Lee, “Pixel art color palette synthesis,” in Information Science and Applications. Springer, 2015, pp. 327–334. [25] LEGO® Brick. https://brickhub.org. 描述 碩士
國立政治大學
資訊科學系
110753159資料來源 http://thesis.lib.nccu.edu.tw/record/#G0110753159 資料類型 thesis dc.contributor.advisor 紀明德 zh_TW dc.contributor.advisor Chi, Ming-Te en_US dc.contributor.author (Authors) 黃威 zh_TW dc.contributor.author (Authors) Huang, Wei en_US dc.creator (作者) 黃威 zh_TW dc.creator (作者) Huang, Wei en_US dc.date (日期) 2024 en_US dc.date.accessioned 1-Mar-2024 13:42:18 (UTC+8) - dc.date.available 1-Mar-2024 13:42:18 (UTC+8) - dc.date.issued (上傳時間) 1-Mar-2024 13:42:18 (UTC+8) - dc.identifier (Other Identifiers) G0110753159 en_US dc.identifier.uri (URI) https://nccur.lib.nccu.edu.tw/handle/140.119/150171 - dc.description (描述) 碩士 zh_TW dc.description (描述) 國立政治大學 zh_TW dc.description (描述) 資訊科學系 zh_TW dc.description (描述) 110753159 zh_TW dc.description.abstract (摘要) 樂高公司以積木的多樣性深受大人和小孩喜愛，隨著模型複雜度 的增加，人們對樂高模型的組裝有了更高的要求。以樂高浮雕系列為 例，模型以其高度立體的設計和細緻的細節而聞名，處理複雜的三維 空間和結構問題上的能力，使組裝過程更具挑戰性。 本研究著重在處理樂高浮雕系列的複雜性。在這一過程中，我們 需要克服積木的幾何形狀、分層架構和結構強度等多重挑戰，同時必 須在有限的樂高磚資源下實現豐富多樣的創意。為了解決這些問題， 我們採用了三項關鍵技術：圖像分層、樂高生成技術與相似度量化分 析。首先，透過圖像分層技術，我們得以細緻地將輸入圖像分為前景 和背景，深入切分圖像中的細節，進而突顯更多層次的圖像細節。其 次，我們應用樂高生成技術，在區域內最大化平鋪樂高磚，確保模型 的結構穩固，同時解決超像素問題。最後，我們運用相似度量化分析 演算法來比較生成的模型和原始輸入圖像的相似度，全面評估和比較 各種模型的表現。這項分析不僅確保了模型的忠實還原，同時也為我 們提供了改進的空間，以進一步提高模型的精確度和真實感。這些技 術的綜合應用為樂高浮雕系列的設計提供了全新的方法和解決方案， 進一步滿足了樂高愛好者的多樣性。 zh_TW dc.description.abstract (摘要) The LEGO company’s diverse building blocks are loved by both adults and children. As models become more complex, there are higher demands for assembling LEGO models. For example, the LEGO relief series, known for its intricate three-dimensional design , presents challenges in handling complex spatial and structural issues. Our study focuses on addressing the complexity of the LEGO relief series. We employ three key technologies: image segmentation, LEGO generation techniques, and similarity quantification analysis. Image segmentation divides input images into foreground and background, emphasizing more layers of detail. LEGO generation techniques maximize brick placement for structural stability while solving the superpixel problem. Similarity quantification analysis ensures faithful reproduction of models and provides room for improvement.By applying these technologies, we offer new methods and solutions for designing LEGO relief series, catering to the diverse interests of LEGO enthusiasts. en_US dc.description.tableofcontents 致謝 i 摘要 ii Abstract iii 目錄 iv 圖目錄 vi 第一章 緒論 1 1.1 研究動機與目的 1 1.2 問題描述 3 1.3 論文貢獻 4 第二章 相關研究 5 2.1 平鋪問題 5 2.2 分層架構 6 2.3 超像素問題 6 2.4 最佳化 7 第三章 研究方法與步驟 9 3.1 資料預處理 10 3.1.1 分層 (Layer) 10 3.1.2 額外分層架構 14 3.1.3 資料預處理 14 3.2 風格化 15 3.3 平鋪問題 16 3.4 顏色對應 16 3.4.1 顏色量化分析 19 3.5 量化分析 21 3.6 分治法 (divided-and-conquer) 22 第四章 實驗與結果 24 4.1 實驗數據比較 25 4.1.1 不同 ColorMapping 方法比較 25 4.1.2 K 值選擇與比較 26 4.1.3 組裝結果大小 27 4.1.4 圖形位移 28 4.2 結果 28 4.3 限制 36 第五章 結論與未來展望 39 5.1 結論 39 5.2 未來展望 40 附錄 A 41 參考文獻 43 zh_TW dc.format.extent 19560266 bytes - dc.format.mimetype application/pdf - dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0110753159 en_US dc.subject (關鍵詞) 樂高 zh_TW dc.subject (關鍵詞) 圖神經網路 zh_TW dc.subject (關鍵詞) 超像素問題 zh_TW dc.subject (關鍵詞) LEGO en_US dc.subject (關鍵詞) Graphic neural network, en_US dc.subject (關鍵詞) Superpixel en_US dc.title (題名) 以圖神經網路將 2.5D 樂高建構映射至平鋪問題之方法 zh_TW dc.title (題名) Mapping 2.5D Lego Construction into Tiling Problem with Graph Neural Network en_US dc.type (資料類型) thesis en_US dc.relation.reference (參考文獻) [1] LEGO® Starry Night. https://www.lego.com/zh-tw/categories/adults-welcome/ article/details-of-van-gogh-starry-night. [2] LEGO® Great Wave. https://www.lego.com/zh-tw/categories/adults-welcome/ article/how-we-made-the-lego-great-wave. [3] LEGO® Wind God and Thunder God Screens. https://toymim.com/review/ lego-store-narita-airport-report-2020-01. [4] A. Rivers, T. Igarashi, and F. Durand, “2.5 d cartoon models,” ACM Transactions on Graphics (TOG), vol. 29, no. 4, pp. 1–7, 2010. [5] H. Xu, K. H. Hui, C.-W. Fu, and H. Zhang, “Tilingnn: learning to tile with selfsupervised graph neural network,” arXiv preprint arXiv:2007.02278, 2020. [6] LEGO® Brick Modified . https://rebrickable.com/parts/87087/ brick-special-1-x-1-with-stud-on-1-side/. [7] R. Ranftl, K. Lasinger, D. Hafner, K. Schindler, and V. Koltun, “Towards robust monocular depth estimation: Mixing datasets for zero-shot cross-dataset transfer,” IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2020. 43 [8] R. Ranftl, A. Bochkovskiy, and V. Koltun, “Vision transformers for dense prediction,” ArXiv preprint, 2021. [9] Z. Wu, S. Pan, F. Chen, G. Long, C. Zhang, and S. Y. Philip, “A comprehensive survey on graph neural networks,” IEEE transactions on neural networks and learning systems, vol. 32, no. 1, pp. 4–24, 2020. [10] L. Sacht, “Structure-aware bottle cap art,” Computers & Graphics, vol. 107, pp. 277–288, 2022. [11] J. Allebach and P. W. Wong, “Edge-directed interpolation,” in Proceedings of 3rd IEEE International Conference on Image Processing, vol. 3. IEEE, 1996, pp. 707–710. [12] R. E. Carlson and F. N. Fritsch, “Monotone piecewise bicubic interpolation,” SIAM journal on numerical analysis, vol. 22, no. 2, pp. 386–400, 1985. [13] 翁瑋辰, “具樂高平滑化之影像樂高風格化技術,” 2019. [14] K. He, G. Gkioxari, P. Dollár, and R. Girshick, “Mask r-cnn,” in Proceedings of the IEEE international conference on computer vision, 2017, pp. 2961–2969. [15] 王祥宇, “以圖神經網路將二維樂高建構映射至平鋪問題之方法,” 2022. [16] P. Lei, S. Xu, and S. Zhang, “An art-oriented pixelation method for cartoon images,” The Visual Computer, pp. 1–13, 2023. [17] R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 586–595. 44 [18] R. Gower, A. Heydtmann, and H. Petersen, “Lego: Automated model construction,” 1998. [19] M.-H. Kuo, Y.-E. Lin, H.-K. Chu, R.-R. Lee, and Y.-L. Yang, “Pixel2brick: Constructing brick sculptures from pixel art,” in Computer Graphics Forum, vol. 34, no. 7. Wiley Online Library, 2015, pp. 339–348. [20] S.-J. Luo, Y. Yue, C.-K. Huang, Y.-H. Chung, S. Imai, T. Nishita, and B.-Y. Chen, “Legolization: Optimizing lego designs,” ACM Transactions on Graphics (TOG), vol. 34, no. 6, pp. 1–12, 2015. [21] H. Xu, K.-H. Hui, C.-W. Fu, and H. Zhang, “Computational lego technic design,” arXiv preprint arXiv:2007.02245, 2020. [22] K. Lennon, K. Fransen, A. O’Brien, Y. Cao, M. Beveridge, Y. Arefeen, N. Singh, and I. Drori, “Image2lego: customized lego set generation from images,” arXiv preprint arXiv:2108.08477, 2021. [23] Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, “Image quality assessment: from error visibility to structural similarity,” IEEE transactions on image processing, vol. 13, no. 4, pp. 600–612, 2004. [24] M.-R. Huang and R.-R. Lee, “Pixel art color palette synthesis,” in Information Science and Applications. Springer, 2015, pp. 327–334. [25] LEGO® Brick. https://brickhub.org. zh_TW
