License plate recognition based on machine learning with circular shape histogram

In this study, a license plate recognition method is presented by constructing feature vectors of license plate characters with circular shape histogram. Firstly, plate recognition methods in the literature, property extraction methods of plate characters and classification methods of characters in line with these methods were introduced. Then, circular shape histogram models of digital image histogram types are introduced. There are some preliminary operations that must be performed on the plate images in order to generate feature vectors of the plate characters with circular shape histogram. Accordingly, the noise on the plate images were reduced and segmentation and thresholding operations were performed in order to examine the characters as different objects. In order to obtain the feature vectors of the characters, the most suitable model was determined from the circular shape histogram models. The obtained feature vectors were classified by two different methods. The characters were determined by linear method and the method based on machine learning with artificial neural network in which statistical similarities of feature vectors were measured. In this way, feature extraction method with less number of elements was developed compared to existing feature extraction methods.