Á¤º¸°úÇÐȸ ³í¹®Áö C : ÄÄÇ»ÆÃÀÇ ½ÇÁ¦
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
°øºÐ»ê ¹× HOG ±â¼úÀÚ¸¦ »ç¿ëÇÑ ¸ð¾ç º¯È 󸮸¦ À§ÇÑ °ÀÎÇÑ ÀÚµ¿Â÷ ¹øÈ£ÆÇ °ËÃâ±â |
¿µ¹®Á¦¸ñ(English Title) |
Robust Car License Plate Detector for Shape Variations Processing using Covariance and HOG Descriptor |
ÀúÀÚ(Author) |
À±Á¾¹Î
±è´ëÁø
Jongmin Yoon
Daijin Kim
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 18 NO. 10 PP. 0728 ~ 0732 (2012. 10) |
Çѱ۳»¿ë (Korean Abstract) |
º» ³í¹®¿¡¼´Â ½ÇÁ¦ ·Îµåºä¿µ»ó¿¡ Àû¿ëµÉ ¼ö ÀÖ´Â »çÀÌÁî, ºí·¯¸µ, ½ÃÁ¡ º¯È, ȸÀü µî¿¡ °ÀÎÇÑ »õ·Î¿î ¹øÈ£ÆÇ °ËÃâ ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. 4´Ü°èÀÇ ÄɽºÄÉÀ̵带 µµÀÔÇÏ¿© ¿À°ËÃâÀ» ÁÙÀÌ°í, °øºÐ»ê ±â¼úÀÚ¿¡ À§Ä¡Á¤º¸¸¦ °áÇÕÇØ ½Å°æ¸Á¿¡ Àû¿ë½ÃÄ×À¸¸ç, ¸¶Áö¸· ½ºÅ×ÀÌÁö¿¡¼´Â HOG ±â¼úÀÚ¿Í LDA¸¦ »ç¿ëÇØ ÁÖ¾îÁø ¿µ¿ªÀÇ ¹øÈ£ÆÇ ¿©ºÎ¸¦ ÆÇ´ÜÇÏ¿´´Ù. ±× °á°ú, ±Ø½ÉÇÑ È¯°æ º¯È°¡ Á¸ÀçÇÏ´Â DB»ó¿¡¼ 93.5%ÀÇ ³ôÀº °ËÃâÀ²À» °¡Áö¸é¼ 8.3・10-7ÀÇ FPPWÀÇ ³·Àº ¿À°ËÃâÀ²À» °®´Â ¹øÈ£ÆÇ °ËÃâ±â¸¦ ¸¸µé ¼ö ÀÖ¾ú´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
This paper presents a novel method that can detect license plates which have several variations including perspective distortion, size variation, and blurring. We used spatial combinations of covariance descriptors in different positions and a Histogram of Oriented Gradient descriptor as input features. We used a feed-forward nework as a classifier and Linear Discriminant Analysis for verification. To overcome our detector¡¯s high computational cost, we used GPU accelerating technology. This method achieved a 93.5% detection rate while maintaining 8.3・10-7 False Positives Per Window in road view images.
|
Å°¿öµå(Keyword) |
¹øÈ£ÆÇ °ËÃâ±â
°øºÐ»êµð½ºÅ©¸³ÅÍ
Car license plate detection
HOG
Region covariance
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|