Development of the system for recognition of car license plate using artificial intelligence

Abstract

The article proposes a technology for creating a license plate recognition system using artificial intelligence. The license plate recognition program is developed in the PyCharm integrated development environment (IDE), which supports Python and provides access to libraries for machine learning and image processing. Libraries for working with images and text, such as OpenCV, imutils, pytesseract, and others, are installed through the PyCharm terminal. For the processing and recognition of license plates, a set of images containing various car license plates was prepared. These images are made from different angles, with different lighting and quality, which is caused by the need to ensure the appropriate response of the program to different input data to ensure its correct operation. After loading the image using the OpenCV library, it is converted to grayscale, since contour detection and image processing are much easier in grayscale. The Canny algorithm was used to detect contours. Contours in the image are used for further identification of the license plate. Determination of horizontal and vertical contours was performed using the Sobel operator. We use Tesseract OCR to recognize license plate text. After the license plate has been recognized, we compare it with those in the database. After setting up the program, it is necessary to test it with different images to make sure the accuracy of license plate recognition. This allows you to detect possible problems in the recognition process, such as low contrast of the image, contamination of license plates or unstable lighting, tilting or distortion of license plates in the image, which can affect the accuracy of recognition