Vehicle registration plate recognition system using template matching

Registration plate recognition is widely used in detecting speedy cars, traffic law enforcement and electronic toll collection. The problems associated with registration plate recognition are, plate images have different quality, illumination, view angle, distance, complex background and fonts. To address these problems, image processing tools are used for extracting only the region of interest. Then different luminance and background are removed by changing colored image to grayscale and then in to binary matrix form. Character recognition algorithm is applied on the binary images. All components of image are compared with pre-defined standard template and characters are recognized based on best match. Each matched character is displayed and stored in a log file. This algorithm is tested with vehicle images of different backgrounds and illumination. The camera focus, viewing plane and the distance from the vehicle were varied. The results of experiments are fascinating with good accuracy rate.