A Pulse-Coupled Neural Network approach for image segmentation and its pattern recognition application

This paper presents a model of Pulse-Coupled Neural Network (PCNN) for multispectral image segmentation. Its application for license plate recognition (LPR) is considered; this consists of three processing steps. First step extracts the license plate coordinates from the original image; second step is the PCNN-based segmentation method to obtain a binary image containing only the characters of the license plate; third step consists of a SVM pattern recognition technique applied to each extracted letter. The LPR application has been tested on a big database, containing 200,000 license plate samples. The experimental results concerning recognition score prove the obvious advantage of the PCNN image segmentation model by comparison to the image segmentation techniques based on SOM and Mean Shift.