Automatic localize Iraqi license plate

• Main Authors: Nidhal, K. Ahmed, Al-Saadi, Enas Hamood, Dheyaa, N. Ammar, Hassoon, Oday Obaid
• Format: Conference Proceeding
• Language: English
• Subjects: Algorithms; Corners; Histograms; Image contrast; Image enhancement; Intelligent transportation systems; License plates; Licenses; Mathematical morphology; Object recognition; Preprocessing; Vehicle identification
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0067042

A car license plate locator and recognition have gained much interest in recent years in Intelligent Transportation System research. The license plate locator is an important stage in the license plate recognition process. In this paper, a new technique to automatically locate the license plate has been proposed, and then recognize it. The proposed technique is divided into two stages, stage one is to locate and corrupt the license plate, and the second stage focuses on the recognition of the license plate. In the first stage image preprocessing (denoising, convert image to binary image, and detect edges) is implemented, then recognizes the region that may contain the license plate, this is done by finding the connected regions using the mathematical morphology (small and very large regions removed). Then the corners in the image are extracted by using the Harris algorithm. Finally, verifies which one of the nominated regions is a license plate, this is done by eliminating all regions that have less than two separate corners. The license plate is determined according to the candidate region size and vertical/horizontal histogram. While in the second stage the corrupt plate is preprocessing (convert to gray image, enhance contrast, and remove noise). The license plate image converted to the binary image prior to extract the numbers and recognize them based on the number of pixels for each number. The accuracy of the locating plate is up to 96%. And the accuracy of recognizing the number is up to 100%.