Recognition of characters in plates of private Brazilian vehicles using neural networks

This work considers the recognition of characters in plates of private Brazilian vehicles using neural networks. Initially it defines a standard for the image as a prerequisite for the pre-processing. Afterwards, the reduction of the dimensionality is made in order to make use of the network. For this the paper establishes a figure of merit Q characterizing, amongst some algorithms of character splitting, the best one. Q is based on normalized cross correlation of the desired character image and the one obtained. The pre-processing allowed image improvement, reducing the background noise and increasing the contrast by applying a non-linear function. In the last stage, the neural network recognizes each character of the plate. Preliminary results are shown