µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)
ÇѱÛÁ¦¸ñ(Korean Title) |
3D-IWGANÀ» ÀÌ¿ëÇÑ Ä§Åõ¼º Æ÷ÀåÀçÀÇ 3Â÷¿ø ¹Ì¼¼±¸Á¶ À籸¼º |
¿µ¹®Á¦¸ñ(English Title) |
3D microstructure reconstruction of Permeable Pavement using 3D-IWGAN |
ÀúÀÚ(Author) |
·çµð¾Æ ¿¡Ä« Æ丮
±ÇÁØÈ£
¾ÈÀçÈÆ
Ludia Eka Feri
Joonho Kwon
Jaehun Ahn
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 34 NO. 02 PP. 0022 ~ 0033 (2018. 08) |
Çѱ۳»¿ë (Korean Abstract) |
°¡»ó ½ÇÇèÀº Àç·á Ư¼ºÀ» Æò°¡ÇÏ°í ´Ù»ó Àç·áÀÇ ±â°èÀû °Åµ¿À» ¸ðµ¨¸µÇÏ´Â È¿À²ÀûÀÎ ¹æ¹ýÀÌ´Ù. °¡»ó ½ÇÇè¿¡¼ °¡Áß Áß¿äÇÑ ºÎºÐÀº ¿ø·¡ÀÇ Æ¯¼ºÀ» º¸Á¸ÇÏ¸é¼ ½ÇÁ¦ °³Ã¼¸¦ ´ëüÇÏ´Â ¸ðµ¨ »ý¼º ºÎºÐÀÌ´Ù. 3Â÷¿ø ¹Ì¼¼±¸Á¶ À籸¼ºÀ» À§ÇÑ ÀÌÀüÀÇ ¿¬±¸µéÀº ÁÖ·Î Åë°èÀû Á¢±Ù ¹æ¹ý°ú È®·üÀû ±â¹ý¿¡ ±â¹ÝÀ» µÎ°í ÀÖ´Ù. ÀÌ ¹æ¹ýµéÀº Àç·á¿¡ ´ëÇÑ ¹Ì¸® Á¤ÀÇµÈ Áö½ÄÀ» ¾î´À Á¤µµ ÇÊ¿ä·Î ÇÑ´Ù. ÀÌ ³í¹®¿¡¼´Â, »çÀü Áö½Ä ¾øÀÌ 3Â÷¿ø ¹Ì¼¼ ±¸Á¶¸¦ »ý¼ºÇÏ´Â µö·¯´× ±â¹ÝÀÇ ±â¹ýÀ» Á¦½ÃÇÑ´Ù. Á¦¾ÈÇÑ ½Ã½ºÅÛÀº 3D-IWGAN ¹æ¹ýÀ» »ç¿ëÇÏ¿© ħÅõ¼º Æ÷ÀåÀç 2Â÷¿ø À̹ÌÁö·ÎºÎÅÍ 3Â÷¿ø ¹Ì¼¼ ±¸Á¶ ¸ðµ¨À» »ý¼ºÇÑ´Ù. ½ÇÇè °á°ú¸¦ ÅëÇÏ¿© Á¦¾ÈÇÑ ½Ã½ºÅÛÀÇ ½ÇÇà °¡´É¼ºÀ» È®ÀÎÇÏ¿´´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
A virtual experiment is an efficient way to assess material properties and to model mechanical behavior in multiphase material. The most crucial part of the virtual experiment is a model generation which replaces the real object by preserving its original properties. Previous researches on the 3D microstructure reconstruction are mainly based on statistical approaches and stochastic techniques. These methods require pre-defined knowledge about the materials. Thus, we propose a deep learning method to address the issue. Our system utilizes a 3D-IWGAN method to generate the 3D microstructure model from 2D images of a permeable pavement. Experimental results demonstrate the feasibility of our system to reconstruct a 3D microstructure model for permeable pavement.
|
Å°¿öµå(Keyword) |
»ý¼ºÀû Àû´ë ½Å°æ¸Á
¹Ì¼¼ ±¸Á¶ ºÐ¼®
¹Ì¼¼ ±¸Á¶ 3D ¸ðµ¨
generative adversarial network
microstructure analysis
3D microstructure model
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|