• Àüü
  • ÀüÀÚ/Àü±â
  • Åë½Å
  • ÄÄÇ»ÅÍ
´Ý±â

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ÇÐȸÁö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ÇÐȸÁö > µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

Current Result Document : 8 / 14 ÀÌÀü°Ç ÀÌÀü°Ç   ´ÙÀ½°Ç ´ÙÀ½°Ç

ÇѱÛÁ¦¸ñ(Korean Title) ¹«µå ¹× ÁÖÁ¦ ±â¹Ý À½¾Ç°ú À̹ÌÁö ¸ÅĪ ±â¹ý
¿µ¹®Á¦¸ñ(English Title) Music and image matching scheme based on mood and theme
ÀúÀÚ(Author) ÇÑ»óÈÆ   ±è¿ë¼º   ȲÀÎÁØ   Sanghoon Han   Yongsung Kim   Eenjun Hwang  
¿ø¹®¼ö·Ïó(Citation) VOL 33 NO. 02 PP. 0077 ~ 0088 (2017. 08)
Çѱ۳»¿ë
(Korean Abstract)
½º¸¶Æ® µð¹ÙÀ̽ºÀÇ ¹ßÀü°ú º¸±ÞÀ¸·Î »ç¿ëÀÚµéÀº ¾ðÁ¦ ¾îµð¼­³ª µ¿¿µ»óÀ̳ª À̹ÌÁö, À½¾Ç°ú °°Àº ´Ù¾çÇÑ ¸ÖƼ¹Ìµð¾î ÄÜÅÙÃ÷¸¦ Æí¸®ÇÏ°Ô Á¢±ÙÇÒ ¼ö ÀÖ°Ô µÇ¾ú´Ù. ÇÑÆí, ÄÜÅÙÃ÷ÀÇ ¾çÀÌ Æø¹ßÀûÀ¸·Î Áõ°¡ÇÔ¿¡ µû¶ó »ç¿ëÀÚ¿¡°Ô ÀûÀýÇÑ ÄÜÅÙÃ÷¸¦ È¿°úÀûÀ¸·Î Á¦°øÇÏ´Â ¹æ¹ýµéÀÌ ¿ä±¸µÇ¾ú°í À̸¦ À§ÇÑ ´Ù¾çÇÑ ¿¬±¸µéÀÌ ÁøÇàµÇ¾ú´Ù. ƯÈ÷, À½¾ÇÀ̳ª À̹ÌÁö¿Í °°Àº ÄÜÅÙÃ÷ÀÇ °æ¿ì, ±×µéÀÌ °¡Áö´Â °¨¼ºÀûÀÎ Á¤º¸°¡ »ç¿ëÀÚÀÇ ÄÜÅÙÃ÷¿¡ ´ëÇÑ ¸¸Á·µµ¸¦ ³ôÀÌ°í ÀÖ¾î °Ë»öÀ̳ª Ãßõ ºÐ¾ß¿¡¼­ À¯¿ëÇÏ°Ô »ç¿ëµÇ°í ÀÖ´Ù. ´õ¿íÀÌ Àü½ÃÀåÀ̳ª ¿£ÅÍÅ×ÀθÕÆ® ºÐ¾ßÀÇ °æ¿ì À½¾Ç°ú À̹ÌÁö¸¦ µ¿½Ã¿¡ Á¦°øÇÏ¿© ÄÜÅÙÃ÷¿¡ ´ëÇÑ »ç¿ëÀÚ ¸ôÀÔµµ¿Í ¸¸Á·µµ¸¦ Çâ»ó½ÃÅ°°íÀÚ ÇÏ¿´À¸¸ç, À̸¦ À§ÇÑ À½¾Ç°ú À̹ÌÁöÀÇ ¿¬°ü¼º °è»êÀÌ Áß¿äÇÑ À̽´°¡ µÇ¾ú´Ù. º» ³í¹®¿¡¼­´Â À½¾Ç°ú À̹ÌÁö°¡ °¡Áö´Â ¹«µå¿Í ÁÖÁ¦¸¦ ÃßÃâÇÏ°í À̸¦ ±â¹ÝÀ¸·Î À¯»çÇÑ ÄÜÅÙÃ÷¸¦ È¿°úÀûÀ¸·Î »ç¿ëÀÚ¿¡°Ô Á¦°øÇÏ´Â ±â¹ýÀ» Á¦¾ÈÇÑ´Ù. ±¸Ã¼ÀûÀ¸·Î, À̹ÌÁö ¹«µå ÀνÄÀ» À§Çؼ­´Â À̹ÌÁö¿¡ Æ÷ÇԵǾî ÀÖ´Â »ç¹°µé°ú ±×µéÀÇ »ö»ó Ư¡À» ÃßÃâÇÏ°í, À½¾Ç ¹«µå ÀνÄÀ» À§Çؼ­´Â À½¾Ç¿¡¼­ »ç¶÷µéÀÌ µè°í Á¤ÀÇÇÑ ¹«µå Ư¡°ú À½¾ÇÀÇ °¡»ç Ư¡À» ÀÌ¿ëÇÑ´Ù. ÃßÃâÇÑ À½¾Ç°ú À̹ÌÁöÀÇ ¹«µå Ư¡µéÀ» AV(Arousal-Valence) °ªÀ¸·Î º¯È¯ÇÏ°í AV ¸ðµ¨ À§¿¡ ¸ÊÇÎÇÏ°í, ¸ðµ¨ »ó¿¡¼­ÀÇ °Å¸®¸¦ °è»êÇÏ¿© ¹«µå À¯»çµµ·Î »ç¿ëÇÏ¿´´Ù. ÇÑÆí, À½¾Ç°ú À̹ÌÁöÀÇ ÁÖÁ¦¸¦ ÆľÇÇϱâ À§Çؼ­ À̹ÌÁö³ª À½¾Ç °¡»ç¿¡¼­ ¾ð±ÞµÈ »ç¹°µéÀ» ¹ÙÅÁÀ¸·Î ÁÖÁ¦ ¿ÂÅç·ÎÁö¸¦ ±¸¼ºÇÏ¿´´Ù. ¸¶Áö¸·À¸·Î ÄÜÅÙÃ÷ÀÇ ¹«µå¿Í ÁÖÁ¦¸¦ ±â¹ÝÀ¸·Î À½¾Ç°ú À̹ÌÁö¸¦ ¸ÅĪÇÏ´Â ½ÇÇèÀ» ÅëÇÏ¿© Á¦¾ÈÇÏ´Â ±â¹ýÀÇ ¼º´ÉÀ» º¸ÀδÙ.
¿µ¹®³»¿ë
(English Abstract)
Various smart devices have enabled users to conveniently access diverse multimedia contents, such as videos, images, and music, anywhere and anytime. Meanwhile, as the volume of contents has become increased explosively, effective methods for providing appropriate contents to users have been required and many works have been done to that purpose. In particular, for the contents such as music and images, emotional information has been used popularly in search or recommendation because such emotional information is known to increase user satisfaction. Moreover, in the exhibition and entertainment field, music and image are presented at the same time to increase the immersion and satisfaction of users. Therefore, relevance calculation of music and images has became an important issue. In this paper, we propose a method to extract the mood and theme of music and images and provides contents to users based on their mood and theme similarity. More specifically, for image mood recognition, we extract the objects in the image and their color features. In order to recognize the music mood, we use the mood characteristic that people defined by listening and lyric characteristic. The mood features of music and image are converted into AV(Arousal-Valence) values, mapped onto the AV plane, and the distance on the plane is defined as the mood similarity of the contents. On the other hand, in order to grasp the theme of music and image, we have constructed a theme ontology based on the objects in image and music lyrics. Finally, we show the performance of our proposed scheme that matches music and image based on mood and theme of contents through experiments.
Å°¿öµå(Keyword) ¹«µå ÀνĠ  Arousal-Valence ¸ðµ¨   À̹ÌÁö À½¾Ç ¸ÅĪ   ÁÖÁ¦ ¿ÂÅç·ÎÁö   Mood recognition   Arousal-Valence model   Image and music matching   Theme ontology  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå