Federated Learning-Based Vehicle Number Plate Recogntion in IoVs

Artificial intelligence is widely used in a variety of industries. AI technology drives much of what we do. In a similar vein, as AI-based technologies advance, smart automobiles and the Smart Transport system will likewise experience revolutionary transformation. Different techniques are applied to create a system that is used to manage traffic and increase security inside the transportation network, different techniques are used. The automatic number recognition system (ANPR) described in this research can extract an image of a vehicle license plate by employing image processing methods. To make things easier, the proposed system may be operated without the installation of any extra GPS-like devices. The suggested system consists of image processing techniques, such as filters to eliminate blur and noise when distantly acquired photographs of moving vehicles are taken. To obtain the region of interest, its edges are detected, and an image is cropped. The procedure for better outcomes includes normalization, localization, image enhancement, restoration, and character retention approaches. Its effectiveness may be negatively impacted by the state of the license plate, unconventional formats, complex vision, camera quality, camera position, tolerance for distortion, motion blur, contrast-related issues, reflections, limitations in a processing unit, environmental factors, indoor/outdoor or time-independent shots, software tools, or other hardware-based restrictions.

Even with the greatest algorithms, a successful ANPR system implementation might need extra computer hardware to boost the proposed System’s accuracy.

Keywords: ANPR, Computer vision, Edge detection, Gaussian blur, OCR, Segmentation.