An elegant approach for automatic license plate recognition

Automatic license plate detection is an active and challenging field of research requiring multi-disciplinary technologies, such as machine vision, pattern recognition and artificial intelligence. This paper presents a simple, elegant and robust approach to automatic recognition of license plate, specifically suited to vehicles in the Indian subcontinent, wherein wide variations in the formats of license plates have to be dealt with. The task entails multilayered stages of pre-processing the image in binary, filtering, morphological processing, and analyzing connected components to extract characters from the license plate. The extracted characters are then passed through a character-detector which subsequently fuses the analysis of correlation through three different correlation metrics to detect the characters within 30 seconds. This approach was tested on 90 manually acquired datasets with an accuracy of over 90%. Thus, this approach achieves accuracy that is comparable to existing methodologies with greater simplicity.