Statistical analysis of the Discrete Fourier Transform for localization in an image

摘要

One of the important steps in the recognition of a license plate in an image is locating the license plate in the captured image. From the localized area, image regions are extracted and recognized as characters, leading to the identification of the vehicle. License plate recognition is a known problem and many contributions have been made to its resolution. Research continues due to the diversity of the license plates and the difficulty of reaching an absolute reliability. We introduce a statistical analysis of the Discrete Fourier Transform of the license plate image data. The method is applied to a color-based histogram thresholded image. The thresholding method improves the binarization of the images and leads the localization method to achieve good results.