在如何調整漫畫成適合手持裝置閱讀的問題中，需要將故事框從漫畫中擷取出來，並對其內容物進行處理，使結果更適合手持裝置上的閱讀，而確定故事框格的位置是最基礎且重要的步驟。但是，由於各系列或各集漫畫中，結構特徵皆可能不相同，例如，各系列漫畫會偏好使用不同的邊框線段長度，使得框格定位的演算法往往需要手動參數調整以達到最高正確率。本論文提出一個利用自動結果評分來找尋最佳參數的自動參數最佳化系統，以減少演算法中的手動調整。此結果評分依據兩個標準。分別為，框格偵測結果的總面積覆蓋率，以及結果框格之間互相平行的框線。透過此結果評分機制，本論文的方法能夠自動產生最佳結果的參數組合。

除了參數的問題之外，目前由其他研究所提出的框格定位演算法也無法處理「破圖」的狀況。所謂破圖是現代新式漫畫常見的表現手法，為一圖片中出現大型前景繪圖物件而遮蔽多個框格的狀況，如海賊王(One Piece)中的人物登場介紹。此狀況會造成框格結構元素的大量損失， 也會造成漫畫頁面中，連通物件標記法產生的物件較不具有明顯的結構規則， 降低了目前已連通物件與切割線為基礎的方法的有效性。為此，本研究提出了一個新的框格偵測的演算法。此方法利用線段偵測與邊角偵測產生本論文中所稱的「構件」，再透過這些構件的互相配對定位出框格的範圍。由於此方法不需使用連通物件來產生切割線或框格範圍。因此，除了一般框格之外，本方法也可以用於處理被破圖狀況中框格的特殊情形。

本論文最後呈現三個實驗，以驗證本論文提出的自動參數最佳化系統能夠找出最合適於一漫畫頁面的最佳參數，與驗證本論文的框格偵測法能夠有效的在破圖狀況中偵測框格位置。在第一個實驗中，本研究將本論文提出之自動參數最佳化系統應用於本論文的框格偵測法，並使用本研究的漫畫圖庫進行實驗，以評估參數最佳化系統對於提升框格偵測成功率的有效性。此實驗也使用本論文的框格偵測法與其他方法處理相同漫畫圖庫，以得到未使用自動參數最佳化系統的比較基準，並在實驗最後呈現了所有方法的結果數據。這些數據顯示了自動參數最佳化系統能夠自動調整參數而產生更正確的結果，因此產生了高於比較基準的正確率。在第二個實驗中，本研究將本論文提出之框格偵測法與其他研究的方法套用於本研究的破圖圖庫，以評估各方法應用於破圖頁面的表現， 並在最後提出得到的結果結果數據。這些數據顯示了由於本研究提出的框格定位演算法利用構件配對而不是單純的連通物件分析，而能達到比其他方法還高的正確率。在第三個實驗中，本研究也對過去方法所使用的實驗資料進行處理，並提出結果比較，顯示了本研究提出的參數最佳化系統的有效性。

In the problem of making manga easier to read on handheld devices, the most essential and basic step is locating positions of each panel on the page since processing on the contents in the area of panels was required to make them more readable on handheld devices. However, due to the varying layout structure features in each series or volume of manga such as the different panel width used in every series, manual parameter adjustment is often needed to achieve the best possible result for many panel extraction algorithms. In this paper, we propose an automatic parameter optimization system based on automatic result evaluation to reduce manual adjustment in panel extraction process. The automatic result evaluation checks for two criteria. The first is the accumulated area of the extracted panels. The second is the number of parallel panel edges between extracted panels. With the automatic result evaluation, our method is able to find correct parameter sets.

Aside from the parameter issues, current algorithms by other studies are not able to process manga pages with large extruding objects. Such pages will make the white gap area between panels very difficult to detect for connected component based algorithms and very hard to distinguish whether the white area is inside or outside of a panel, reducing the effectiveness of connected component component based or division line based algorithm. For this reason, this paper also proposed a novel method for panel extraction. The method employs line segment detection and corner detection to detect what are called "components" in this paper, and match up these components with each other to locate the position of the panels. As the method does not require connected component detection to detect division line or initial panel areas, it can be used to extract panels from both normal manga pages or manga pages with large extruding objects.

At the end of this paper, three experiments are presented to verify that our automatic parameter optimization can detect bad cases and the parameter set with which the algorithm can perform better, and our panel extraction method is able to locate panel positions effectively from manga pages with large extruding objects. The first experiment will compare the results using and without using the automatic parameter optimization to show the optimization is able to improve the result. The second experiment will present the result of applying our method to our collected pages with large extruding objects to show our method is effective for this case. In the third experiment, this study applies the proposed system to the data set used in the past researches and compares results to display our system's effectiveness.

[1] C.-J. Tsai, C.-Y. Yao, P.-y. Chiang, Y.-C. Lai, M.-T. Chi, H.-K. Chu, Y.-S.
Wong, and Y.-S. Wang, “Adaptive manga re-layout on mobile device,” in ACM
SIGGRAPH 2013 Posters, SIGGRAPH ’13, pp. 57:1–57:1, 2013.
[2] Y. Cao, A. B. Chan, and R. W. H. Lau, “Automatic stylistic manga layout,”
ACM Trans. Graph., vol. 31, no. 6, pp. 141:1–141:10, 2012.
[3] X. Pang, Y. Cao, R. W. Lau, and A. B. Chan, “A robust panel extraction
method for manga,” in Proceedings of the 22Nd ACM International Conference
on Multimedia, MM ’14, pp. 1125–1128, 2014.
[4] T. Tanaka, K. Shoji, F. Toyama, and J. Miyamichi, “Layout analysis of treestructured
scene frames in comic images,” in Proceedings of the 20th International
Joint Conference on Artifical Intelligence, IJCAI’07, pp. 2885–2890,
2007.
[5] C. H. Chan, H. Leung, and T. Komura, “Automatic panel extraction of color
comic images,” in Proceedings of the Multimedia 8th Pacific Rim Conference on
Advances in Multimedia Information Processing, PCM’07, pp. 775–784, 2007.
[6] E. Han, K. Kim, H. Yang, and K. Jung, “Frame segmentation used mlp-based
x-y recursive for mobile cartoon content,” in Proceedings of the 12th International
Conference on Human-computer Interaction: Intelligent Multimodal
Interaction Environments, HCI’07, pp. 872–881, 2007.
[7] D. Ishii and H. Watanabe, “A study on frame position detection of digitized
comics images,” in Workshop on Picture Coding and Image Processing,
PCSJ2010/IMPS2010, 2010.
[8] A. K. N. Ho, J. C. Burie, and J. M. Ogier, “Panel and speech balloon extraction
from comic books,” in 2012 10th IAPR International Workshop on Document
Analysis Systems, pp. 424–428, 2012.
[9] H. Samet and M. Tamminen, “Efficient component labeling of images of arbitrary
dimension represented by linear bintrees,” IEEE Trans. Pattern Anal.
Mach. Intell., vol. 10, no. 4, pp. 579–586, 1988.
[10] K. Arai and T. Herman, “Method for automatic e-comic scene frame extraction
for reading comic on mobile devices,” in 2010 Seventh International Conference
on Information Technology: New Generations, pp. 370–375, 2010.
[11] C. Rigaud, N. Tsopze, J.-C. Burie, and J.-M. Ogier, “Robust frame and text extraction
from comic books,” in Proceedings of the 9th International Conference
on Graphics Recognition: New Trends and Challenges, GREC’11, pp. 129–138,
2013.
[12] L. Li, Y. Wang, C. Y. Suen, Z. Tang, and D. Liu, “A tree conditional random
field model for panel detection in comic images,” Pattern Recogn., vol. 48, no. 7,
pp. 2129–2140, 2015.
[13] L. Li, Y. Wang, Z. Tang, and D. Liu, “Comic image understanding based on
polygon detection,” in Document Recognition and Retrieval XX, Proc. SPIE,
pp. 86580B–86580B–11, 2013.
[14] L. Li, Y. Wang, Z. Tang, and L. Gao, “Automatic comic page segmentation
based on polygon detection,” Multimedia Tools Appl., vol. 69, no. 1, pp. 171–
197, 2014.
[15] Y. Wang, Y. Zhou, and Z. Tang, “Comic frame extraction via line segments
combination,” in 2015 13th International Conference on Document Analysis
and Recognition (ICDAR), pp. 856–860, 2015.
[16] NHN Taiwan Corporation, “Comico,” 2017.
[17] Ruian I like Network Co. Ltd., “Manhuaren,” 2017.
[18] W.-T. Chu and H.-H. Wang, “Enabling portable animation browsing by transforming
animations into comics,” in Proceedings of the 2Nd ACM International
Workshop on Interactive Multimedia on Mobile and Portable Devices, IMMPD
’12, pp. 3–8, 2012.
[19] S. Suzuki and K. Abe, “Topological structural analysis of digitized binary images
by border following,” Computer Vision, Graphics, and Image Processing,
vol. 30, no. 1, pp. 32 – 46, 1985.
[20] J. Serra, Image Analysis and Mathematical Morphology. Orlando, FL, USA:
Academic Press, Inc., 1983.
[21] R. Grompone von Gioi, J. Jakubowicz, J.-M. Morel, and G. Randall, “Lsd: a
line segment detector,” Image Processing On Line, vol. 2, pp. 35–55, 2012.
[22] J. Matas, C. Galambos, and J. Kittler, “Robust detection of lines using the
progressive probabilistic hough transform,” CVIU, vol. 78, no. 1, pp. 119–137,
2000.
[23] C. Harris and M. Stephens, “A combined corner and edge detector,” in In Proc.
of Fourth Alvey Vision Conference, pp. 147–151, 1988.
[24] A. G. Khachaturyan, S. V. Semenovskaya, and B. K. Vainshtein, “Statisticalthermodynamic
approach to determination of structure amplitude phases,”
Sov.Phys. Crystallography, vol. 24, no. 5, pp. 519–524, 1979.
[25] R. T. Rockafellar and R. J.-B. Wets, Variational Analysis. Tiergartenstraße 17,
69121 Heidelberg, Germany: Springer-Verlag Berlin Heidelberg, 1998.
[26] K. Arai and T. Herman, “Automatic e-comic content adaptation,” International
Journal of Ubiquitous Computing (IJUC), vol. 1, no. 1, pp. 1 – 11, 2010.
[27] D. Liu, Y. Wang, Z. Tang, L. Li, and L. Gao, “Automatic comic page image
understanding based on edge segment analysis,” in Document Recognition and
Retrieval XXI, Proc. SPIE, pp. 90210J–90210J–12, 2013.
[28] M. Stommel, L. I. Merhej, and M. G. M¨uller, “Segmentation-free detection of
comic panels,” in Proceedings of the 2012 International Conference on Computer
Vision and Graphics, ICCVG’12, pp. 633–640, 2012.