Research and implementation of license plate recognition technology

In order to reduce the impact on recognition speed and accuracy of license plate caused by the image definition differences in condition of all-weather monitoring, a method of license plate recognition for all- weather images is proposed in the paper. In the positioning part, different edge detection operator is selected according to the different image definition, and the license plate texture characteristics are used to remove noises. A corner detecting method which can obtain location information of the candidate region is presented also. For the stroke adhesion problem of Chinese characters, a kind of recognition method on fuzzy outline is presented, which extracts the fuzzy outline of the Chinese character firstly, and then calculates the Fourier descriptors of the fuzzy outline to distinguish characters. This method does not rely on the internal character stroke information, and the Fourier descriptors extracted have invariance on rotation, translation and scale, so it has good robustness. The experimental results show that the positioning rate of all-weather images is about 97.7%, the character recognition rate is about 95.6%, and the plate recognition rate for the whole vehicle is about 93.4%, while the average time is about 0.5 seconds per image. Compared with the traditional method of license plate recognition, the recognition rate is greatly improved in the premise of recognition time being not increased. So it proves that the method is feasible and effective.