2019³â Ãß°èÇмú´ëȸ
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
º´·Ä¿¬»ê ±â¹Ý »ê¾÷¿ë X-¼± computed tomography |
¿µ¹®Á¦¸ñ(English Title) |
Parallel Data Processing for the Industrial X-ray Computed Tomography |
ÀúÀÚ(Author) |
¾çÀ±±â
ÃÖÇü¼®
ÀÌ¿ëÂù
Yoon Gi Yang
Hyeong Seok Choi
Yong Chan Lee
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 23 NO. 02 PP. 0212 ~ 0215 (2019. 10) |
Çѱ۳»¿ë (Korean Abstract) |
º» ¿¬±¸¿¡¼´Â »ê¾÷¿ë ºÎÇ°À» X-¼±À» Åõ°úÇÏ´Â ºñÆı« °Ë»çÀÇ »ê¾÷¿¡¼ ÃÖ±Ù¿¡ °³¹ßµÈ computed tomography ÀåºñÀÇ µ¥ÀÌÅ͸¦ °í¼Ó º´·Äó¸® ¿¬»êÀ» ÅëÇÏ¿© È¿°úÀûÀ¸·Î ó¸®ÇÏ´Â ±â¹ý¿¡ ´ëÇؼ ´Ù·é´Ù.
º» Àåºñ¿¡¼ Ãâ·ÂµÈ 3D CT raw µ¥ÀÌÅÍ´Â ¸Å¿ì ¹æ´ëÇÑ Å©±âÀÇ µ¥ÀÌÅͷμ À̸¦ ÀϹÝÀûÀÎ ÇÁ·Î¼¼¼·Î ó¸®Çϱ⿡´Â ³Ê¹« Å« ºÎ´ãÀÌ µÈ´Ù. µû¶ó¼, º» ¿¬±¸¿¡¼´Â CT µ¥ÀÌÅ͸¦ ÀÏ¹Ý PC·Î Àü´Þ¹Þ¾Æ À̸¦ °í¼Ó ó¸®ÇÒ ¼ö ÀÖ´Â Ç÷§ÆûÀ» ±¸ÃàÇÏ°í, ºñ±³Àû ÀûÀº ºñ¿ëÀ¸·Î CT µ¥ÀÌÅÍÀÇ Ã³¸®¸¦ ÅëÇÏ¿© 3Â÷¿ø CT µ¥ÀÌÅÍÀÇ °¢Á¾ À¯¿ëÇÑ Á¤º¸¸¦ µµÃâÇÒ ¼ö ÀÖ´Â ½Ã½ºÅÛÀ» ±¸ÃàÇÏ¿´´Ù. ½ÇÇè¿¡ »ç¿ëµÈ ÁÖ¿äÇÑ Àü·«Àº Àß ¾Ë·ÁÁø ¿¬»ê ÅøÀÎ MatlabÀÌ ´ÙÁß ¿¬»êÀåÄ¡ÀÎ GPU¸¦ Áö¿øÇϴµ¥ Âø¾È ÇÑ °ÍÀÌ´Ù. ƯÈ÷ º» ¿¬±¸¿¡¼´Â ¿¬»ê·®ÀÌ ¹æ´ëÇÑ voxel (volume elemment) ±â¹ÝÀÇ Æ¯Â¡ ÃßÃâ ¾Ë°í¸®µë°ú °°ÀÌ º¹ÀâÇÏ°í ¹æ´ëÇÑ ¿¬»êÀ» GPU ±â¹Ý ÇÁ·Î¼¼¼¿¡¼ ó¸®ÇÔÀ¸·Î½á, °í°¡ÀÇ Àü¿ë ¼ÒÇÁÆ®¿þ¾î¿¡¼ ÇÏ´ø Àü¹® ±â´ÉÀ» Àú°¡ÀÇ Çϵå¿þ¾î ±â¹Ý¿¡¼ ½ÇÇàÇÒ ¼ö ÀÖ¾ú´Ù. À̸¦ ÅëÇØ »ê¾÷ü¿¡¼ x-¼± ´ÜÃþÃÔ¿µ Àåºñ¸¦ Á» ´õ ½±°Ô »ç¿ëÇÒ ¼ö ÀÖ´Â ±â¹ÝÀ» ±¸ÃàÇÏ¿´´Ù. |
¿µ¹®³»¿ë (English Abstract) |
In this research, some novel parallel processing schemes are implemented for the 3D X-ray industrial CT (computed tomography) which is used to inspect the internal areas of industrial parts without any destruction of the part itself. The data extracted from the CT has much big size so that it spends too much time to process in the general purpose PC environments. Thus, in this research, some novel parallel processing platforms are exploited for the voxel data transfered from the CT with relatively cheap computing environments, which results some meaningful information extraction through voxel data processing. Recently, Matlab and some GPU makers has released much useful tools for the parallel processing applications such as the raw CT data processing. Moreover, high performance GPU processors with relatively low cost products are available nowadays, so we can implement low cost high performance parallel computing machine for the CT data processing. From the research results, many parts manufactueres can use the CT machine more easily than before. |
Å°¿öµå(Keyword) |
Parallel Processing
GPU
Matlab
3D image process
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|