Intelligent License Plate Searching and Content Segmentation in Image Processing

• Main Authors: LIN,TAI-LIANG, 林泰良
• Other Authors: 陳永耀
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/20801088233643083803

碩士 === 國立臺灣大學 === 電機工程學研究所 === 88 === This thesis describes an integrated approach of license plate searching and its content segmentation in image processing. A license plate recognition system can be very useful in parking space management and monitoring. The procedures of license plate recognition include searching of license plate image, its content segmentation, and character recognition. The first two tasks were the focus of this thesis. Intelligent schemes were developed to locate and segment the license plate image from a general surveillance picture. The license plate was located based on the characteristics of high gradient values of pixel gray levels. The image corruption from the scanning interlace problem is solved by interpolations. The inclination of license plates were determined by a contour method and thereby compensated. The contour of the license plate content also provides the key information for segmentations. Overall, an efficient algorithm is developed to achieve an over 90% rate of success in both searching and segmentation stages. The test samples include 1450 images taken in various locations and periods of time.