OPTICAL CHARACTER RECOGNITION NEURAL NETWORK SYSTEM FOR MACHINE-PRINTED CHARACTERS

摘要

Character images to be sent to a neural network trained to recognize a predetermined set of symbols, are first processed using a pre-processor for optical character recognition, which normalizes the character images. The output of the neural network is processed by an optical character recognition post-processor. The post-processor corrects erroneous symbol identifications made by the neural network. The post processor identifies special symbols and unidentifiable symbol breakages by the neural network after normalization of characters. As for the characters identified by the neural network with a low score, the post processor is trying to find and separate the kerned adjacent characters and related characters. The joined characters are separated in a method of nine successively initiated processes depending on the geometric parameters of the image. When all other methods fail, the post-processor selects either the second or the third symbol having the highest score, identified by the neural network based on the likelihood of confusion between the second and the third symbol having the highest score, and the symbol having the highest score.