A vehicle number plate recognition system using region-of-interest based filtering method

Abstract

In this paper, a vehicle number plate recognition (VNPR) system is implemented for Indian vehicles. For this purpose, we propose a ‘region-of-interest (ROI)’-based filtering method to locate the candidate regions of number plate (NP) occurrence. In the proposed filtering method, candidate regions are located in the NP image by detecting vertical edges, removing long edges and stationary regions. Finally, the NP region is segmented from the candidate regions before passing it to the optical character recognition (OCR) system for recognition of characters and digits present in the number plate. The novelty of the proposed VNPR system lies in exploring the ROI-based filtering method which improves the overall performance of the proposed VNPR system. The proposed system has been tested using various NP images of vehicles extracted from real-life video sequences that vary along the dimensions of light, scale and orientation. The experimental results demonstrate the robustness of the proposed method.