2019³â Ãß°èÇмú´ëȸ
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
ÀÚµ¿ÈµÈ ¼ýÀÚ ¼öÈ ÀνÄÀ» À§ÇÑ ÀâÀ½ Á¦°Å ¾Ë°í¸®Áò |
¿µ¹®Á¦¸ñ(English Title) |
Noise Removal Algorithm for Automated Numeric Sign Language Recognition |
ÀúÀÚ(Author) |
À̽Âȯ
À¯Àçõ
SeungHwan Lee
JaeChern Yoo
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 23 NO. 02 PP. 0340 ~ 0342 (2019. 10) |
Çѱ۳»¿ë (Korean Abstract) |
ÀüÀÚ±â±âµéÀÌ °á°ú¹°·Î½á Á¦°øÇÏ°í ÀÖ´Â µ¥ÀÌÅ͵éÀº ÀâÀ½ÀÌ Á¸ÀçÇÑ´Ù. ÀâÀ½Àº ȸ·Î ¹× Åë½Å »óÅ µîÀÇ ÀÌÀ¯·Î ÀÌÀüº¸´Ù ¸¹¾ÆÁú °¡´É¼ºÀÌ ÀÖ°í, ÀÌ·¸°Ô Áõ°¡ÇÑ ÀâÀ½À» Æ÷ÇÔÇÏ´Â µ¥ÀÌÅÍ´Â ¾Ë°í¸®ÁòÀÇ °á°ú°ª¿¡ ¿µÇâÀ» ¹ÌÄ¥ ¼ö ÀÖ´Ù. µû¶ó¼ º» ³í¹®¿¡¼´Â ¼ýÀÚ¸¦ ³ªÅ¸³»´Â ¼öÈ À̹ÌÁö°¡ Æ÷ÇÔÇÏ´Â ¿©·¯ °¡Áö ÀâÀ½À» ¾Ë°í¸®ÁòÀ» ÀÌ¿ëÇÏ¿© Á¦°ÅÇÏ°í, ¼Æ÷Æ® º¤ÅÍ ¸Ó½Å ±â¹ÝÀÇ HOG(Histogram of Oriented Gradients) Ư¡ º¤Å͸¦ »ç¿ëÇÏ¿© ºÐ·ùÇÔÀ¸·Î½á ÀâÀ½À» Á¦°ÅÇÑ µ¥ÀÌÅ͸¦ »ç¿ëÇÑ ¾Ë°í¸®ÁòÀÌ ÀâÀ½À» Æ÷ÇÔÇÑ µ¥ÀÌÅͺ¸´Ù ³ôÀº Á¤È®µµ¸¦ º¸ÀÏ ¼ö ÀÖ´Ù´Â °¡´É¼ºÀ» Á¦¾ÈÇÑ´Ù. |
¿µ¹®³»¿ë (English Abstract) |
The data that electronic devices provide as a result is noisy. Noise is likely to be higher than before for reasons of circuit and communication conditions, and data with this increased noise can affect the algorithm's output. Therefore, in this paper, various noises included in a sign language image representing a number are removed using an algorithm and classified using a HOG (Histogram of Oriented Gradient) feature vector based on a support vector machine. In this study, we proposes the possibility that the algorithm using the noise-free data can show higher accuracy than the data with noise. |
Å°¿öµå(Keyword) |
Noise Removal Algorithm
Support Vector Machine
Sign Language
HOG(Histogram of Oriented Gradients
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|