A System for Locating License-plate Area from Images Taken under Complex Conditions

In this paper, an efficient edge-based approach, including implementation of a hardware system, for locating car license plate area from images taken under relatively complex conditions is presented. The proposed method is divided into three steps: 1) preprocessing - edges are detected from the input image and thresholded, 2) global search - candidate areas are sought from the edge images, and 3) local search - exact plate area is determined based on the structural properties of the plate. For many images collected from a large underground parking place, the accuracy of the global search is 98% when the first and the second candidates are considered. And the local search algorithm correctly locates the plate area for 97% of the images.

• Type: Journal
• Year: 2002
• Where: MVA
• Authors: Sunghoon Kim, Changsub Park, Gyeonghwan Kim
• Keywords: Car License Plate, Computer Vision, Images, MVA 2002, Plate Area