Abstract:

In order to utilize the template matching method more efficiently to implement the recognition of license plate characters, we propose a fuzzy template matching method based on mathematical morphology which  combines the mathematical morphology and the fuzzy set theory. Firstly, for each binary image pixel point and its 8neighborhood, the degree that the center pixel point belongs to the current character can be described by weighting. Secondly, we use a 4×4window to select a representative point and traverse it through the entire character with partially overlapping, and the fuzzy membership matrix of each character image is constructed. Then the Hamming approach degree calculation is utilized to classify the characters to be recognized and the recognition is achieved. Finally, we program the fuzzy template matching method on the Matlab and test the recognition results of actual license plate characters. Compared with the traditional template matching method, the test results show that our proposal can improve the recognition accuracy of license plate characters significantly.

Key words: mathematical morphology;fuzzy set;template matching;license plate characters;Matlab