Recognition and Orientation Invariant Transform for Character Image - Example for License plate

• Main Authors: Huang, Chifu, 黃淇富
• Other Authors: Su, Chingliang
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/3m6s2u

碩士 === 大葉大學 === 工業工程與科技管理學系 === 101 === In recent years the surge of vehicles creates much problem, which include vehicle parking as well as theft and other issues. Vehicle management becomes more and more important. A license plate presents the best information of a vehicle. Thus the recognition of a license plate is very important. As the progress of technology, computer can more efficiently to recognize a license plate. In the past studies, researchers have developed various techniques to recognize a license plate. However, when a car plate is rotated or shifted, the previous technique might have difficulty to recognize them. Therefore, this study proposes a new method to solve this problem.In this study SONY MT11i digital camera is used to capture the images of license plates. A platform with computer vision technology developed by Borland C is used to recognize them. The vision technology includes: (1) color conversion, normalization, noise removing, and character segmentation, (2) locating a character edge, (3) edge thinning, (4) edge sampling, (5) edge distance calculating from the character mass centroid, (6) fast Fourier transform computation, and (7) pattern recognition. In this study, the character feature of each character is stored in a database. For recognize an unknown character in a license plate, the pattern of this character is compared with those individual character pattern, which are stored in the aforementioned database, whereby to recognize this unknown character.