This paper describes the application of mathematical morphology to the problem of extracting the characters from a license plate in order to automatically read it. No apriori assumption on the background is made. The key idea is to use as many features of license plates as needed so as to have a low probability than other structures in the input image fulfill all the requirements. The proposed algorithm works in real time on a personal computer. To achieve this, new morphological operators have been defined and are described in this paper.
展开▼
