A Method of License Plate Chinese Character Recognition Based on Image Quality

This paper presents a method of license plate Chinese character recognition based on image quality for improving the recognition rate of low-resolution characters in license plates. Each subset of license plate Chinese character images with the same fuzzy degree are used to construct a corresponding PCA subspace, then the projection residuals distance between the Chinese character to be recognized and each subspace in the specific subspaces which are selected by character's fuzzy degree is calculated. The class which corresponds to the subspace with the minimum distance is the recognition result. The experimental results show that the proposed method has higher recognition rate and faster recognition speed comparing to other ones.