A Binary Method of License Plate Character Based on Object Space Distribution Feature

Abstract

For the problem of the difficult object-extraction in transition region and noise-effect in binary processes, the vehicle license plate characters binary method based on object space distribution feature is proposed. Considering the spatial distribution of object and background, the method with less false-dismissal and more false-alarm can extract more perfect object shape than conventional method. And then the morphology operation is used to throw away noise-isolated point. Actual vehicle images from traffic station are used to test the performance of different binary methods and the experimental results indicate that our algorithm achieved better binary results than other methods in object shape extraction and noises reduction.