Language：English   Publishing type：Research paper (scientific journal)   Publisher：Applied Soft Computing  

Escher-like tiling attempts to design a tile whose copies cover a plane with no overlaps and no gaps. Deforming a given image shape into a tileable shape while maintaining the original shape as far as possible is a difficult problem. A tileable shape similar to a given image can be generated by an analytical optimization method (AOM) that requires no iterative calculations. However, the generated shape is often non-tileable due to edge self-intersections; moreover, as the method is sensitive to small changes of the input image shape, the input shape must be adjusted manually by much trial and error. To avoid this problem, this paper proposes an Escher-Like Tile design method using Hierarchical OptimizatioN (ELTHON), which divides the tile-design problem and resolves the subproblems by two different methods. The given problem (such as the goal figure shape) is modified by an upper-layer optimizer, and feasible solutions (such as the tileable shapes) are found by a lower-layer optimizer based on AOM. Because the upper layer employs metaheuristics such as a genetic algorithm, which support flexible objective functions and constraints, different types of figure similarity metrics and constraints can be selected to avoid the generation of non-tileable shapes. Experimental results showed that ELTHON designed tiles without violating the constraints, whereas 58.9% of the solutions found by AOM violated the constraints. Additionary, in 16 of 32 figures tested, ELTHON produced tileable shapes with higher similarity to given images compared with AOM, whereas AOM outperformed ELTHON in only two figures.

DOI： 10.1016/j.asoc.2021.107771

Scopus

Language：English   Publishing type：Research paper (scientific journal)   Publisher：Journal of Information Science and Engineering  

This paper proposes a method that deforms a prominent movie or animation character into a tileable shape. Tiling is the act of covering the plane with one or a very few types of figures without overlaps and/or gaps. Although some previous methods can transform a given shape into a tileable shape, they cannot easily move the character into a suitably tileable pose. The proposed method learns the latent feature space that abstracts the target character's silhouettes using a convolutional variational autoencoder, and looks for the poses suitable for tiling by optimization in the latent space. Experimental results showed that the proposed method successfully generated tileable figures of the tested character in various poses, some of which were not included in the training dataset.

DOI： 10.6688/JISE.20210537(3).0005

Scopus

Authorship：Lead author   Language：Japanese   Publishing type：Doctoral thesis  

Language：English   Publishing type：Research paper (international conference proceedings)  

DOI： 10.1109/TAAI48200.2019.8959833

Language：Japanese   Publishing type：Research paper (scientific journal)  

Language：English   Presentation type：Oral presentation (general)  

Language：Japanese   Presentation type：Oral presentation (general)  

Language：Japanese   Presentation type：Oral presentation (general)  

Language：Japanese   Presentation type：Oral presentation (general)  

Language：Japanese   Presentation type：Poster presentation  

Application no：特願2024-049773  Date applied：2024.3

Award type：Award from international society, conference, symposium, etc.  Country：Taiwan, Province of China

Award type：Award from Japanese society, conference, symposium, etc. 

Award type：Award from Japanese society, conference, symposium, etc.