Á¤º¸Ã³¸®ÇÐȸ ³í¹®Áö ¼ÒÇÁÆ®¿þ¾î ¹× µ¥ÀÌÅÍ °øÇÐ
ÇѱÛÁ¦¸ñ(Korean Title) |
Stack-AttentionÀ» ÀÌ¿ëÇÑ È帴ÇÑ ¿µ»ó °È ±â¹ý |
¿µ¹®Á¦¸ñ(English Title) |
Blurred Image Enhancement Techniques Using Stack-Attention |
ÀúÀÚ(Author) |
¹Ú串
À̱¤ÀÏ
Á¶¼®Á¦
Park Chae Rim
Lee Kwang Ill
Cho Seok Je
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 12 NO. 02 PP. 0083 ~ 0090 (2023. 02) |
Çѱ۳»¿ë (Korean Abstract) |
ÄÄÇ»ÅÍ ºñÀü¿¡¼ È帴ÇÑ ¿µ»óÀº ¿µ»ó ÀνķüÀ» ÀúÇϽÃÅ°´Â Áß¿äÇÑ ¿äÀÎÀÌ´Ù. ÀÌ°ÍÀº ÁÖ·Î Ä«¸Þ¶ó°¡ ºÒ¾ÈÁ¤ÇÏ°Ô ÃÊÁ¡À» ¸ÂÃßÁö ¸øÇϰųª, ³ëÃ⠽𣵿¾È Àå¸éÀÇ ¹°Ã¼°¡ ºü¸£°Ô ¿òÁ÷ÀÏ ¶§ ¹ß»ýÇÑ´Ù. È帴ÇÑ ¿µ»óÀº ½Ã°¢Àû Ç°ÁúÀ» Å©°Ô ÀúÇϽÃÄÑ °¡½Ã¼ºÀ» ¾àȽÃÅ°¸ç, ÀÌ·¯ÇÑ Çö»óÀº µðÁöÅÐÄ«¸Þ¶óÀÇ ±â¼úÀÌ Áö¼ÓÀûÀ¸·Î ¹ßÀüÇÏ°í ÀÖÀ½¿¡µµ ºÒ±¸ÇÏ°í ºó¹øÇÏ°Ô ÀϾÙ. º» ³í¹®¿¡¼´Â ÇÕ¼º°ö ½Å°æ¸ÁÀ¸·Î ¼³°èµÈ ½ÉÃþ ¸ÖƼ ÆÐÄ¡ °èÃþ ³×Æ®¿öÅ©(Deep multi patch hierarchical network)¸¦ ±â¹ÝÀ¸·Î ¼öÁ¤µÈ ºôµù ¸ðµâÀ» ´ëüÇÏ¿© ÀÔ·Â ¿µ»óÀÇ µðÅ×ÀÏÀ» Àâ°í ÁÖÀÇ ÁýÁß ±â¹ýÀ» µµÀÔÇÏ¿© È帴ÇÑ ¿µ»ó ¼Ó ¹°Ã¼¿¡ ´ëÇÑ ÃÊÁ¡À» ´Ù¹æ¸éÀ¸·Î ¸ÂÃß¾î ¿µ»óÀ» °ÈÇÑ´Ù. ÀÌ°ÍÀº ¼·Î ´Ù¸¥ ½ºÄÉÀÏ¿¡¼ °¢°¢ÀÇ °¡ÁßÄ¡¸¦ ÃøÁ¤ ¹× ºÎ¿©ÇÏ¿© È帲ÀÇ º¯È¸¦ Â÷º°ÀûÀ¸·Î ó¸®ÇÏ°í ¿µ»óÀÇ °ÅÄ£ ¼öÁØ¿¡¼ ¹Ì¼¼ÇÑ ¼öÁرîÁö ¼øÂ÷ÀûÀ¸·Î º¹¿øÇÏ¿© ±Û·Î¹úÇÑ ¿µ¿ª°ú ·ÎÄà ¿µ¿ª ¸ðµÎ Á¶Á¤ÇÑ´Ù. ÀÌ·¯ÇÑ °úÁ¤À» ÅëÇØ ÀúÇÏµÈ ÈÁúÀ» º¹±¸ÇÏ°í È¿À²ÀûÀÎ °´Ã¼ ÀÎ½Ä ¹× Ư¡À» ÃßÃâÇÏ¸ç »ö Ç×»ó¼ºÀ» º¸¿ÏÇÏ´Â ¿ì¼öÇÑ °á°ú¸¦ º¸¿©ÁØ´Ù. |
¿µ¹®³»¿ë (English Abstract) |
Blurred image is an important factor in lowering image recognition rates in Computer vision. This mainly occurs when the camera is unstablely out of focus or the object in the scene moves quickly during the exposure time. Blurred images greatly degrade visual quality, weakening visibility, and this phenomenon occurs frequently despite the continuous development digital camera technology. In this paper, it replace the modified building module based on the Deep multi-patch neural network designed with convolution neural networks to capture details of input images and Attention techniques to focus on objects in blurred images in many ways and strengthen the image. It measures and assigns each weight at different scales to differentiate the blurring of change and restores from rough to fine levels of the image to adjust both global and local region sequentially. Through this method, it show excellent results that recover degraded image quality, extract efficient object detection and features, and complement color constancy. |
Å°¿öµå(Keyword) |
µðºí·¯¸µ
ÁÖÀÇ ÁýÁß ±â¹ý
·¹Æ¼³Ø½º
»ö Ç×»ó¼º
ÄÄÇ»ÅÍ ºñÀü
Deblurring
Attention
Retinex
Color Constancy
Computer Vision
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|