To provide a license plate number recognition device capable of extensive recognition with one television camera.
A series of numbers are segmented from an image taken with a relatively low resolution of about 3 mm/pixel in process 15 to process 17, and a visual field size of imaging is calculated from the position coordinates of the numbers in process 30. The input image of a plate area is enlarged so that the size has a relatively high resolution of about 2mm/pixel in process 32, and a series of numbers are cut out again from this image in process 33. Thereafter, the positions of various codes (motor vehicle official branch code, model code, use code) are determined from the coordinates of the segmented numbers followed by character recognition to recognize a vehicle license plate number. The plate part is segmented, and its image is enlarged to a predetermined size in this way, whereby the area coordinates of all characters can be precisely determined, and the visual field width can be consequently extended to a resolution of about 3 mm/pixel. Therefore, a road can be extensively monitored by use of an inexpensive camera.
COPYRIGHT: (C)2005,JPO&NCIPI
Yoshiyuki Sato
Shuji Omori
Naoko Yoko
Ryusuke Kawatani
株式会社日立情報制御ソリューションズ
JP2001273461A | ||||
JP2001118032A | ||||
JP8190689A | ||||
JP62061181A |
Next Patent: SUPPORT BRACKET ELASTICALLY CONNECTING VEHICLE PART, PARTICULARLY CONNECTING SUB-FRAME TO BODY