In this article we present the implementation of a neural network model trained with a high noise level using a backpropagation algorithm and the experimental results for printed character recognition, based on the idea of using the primary information by reorganising it in a different format. The values obtained at the outputs of each network are processed by using analysis algorithms designed for this purpose. The suggested model is made up of two neural networks and two analysis modules. In M1 Module we designed a value analysis algorithm for all the outputs of the two neural networks in order to select the best values provided by the networks. The M2 Module also contains a designed algorithm, which assesses the data based on the fact that the highest values are directly correlated with the probability of correctly identifying the characters entered into the networks. Results are obtained for noise of up to 50% applied to the input data. The values obtained at the outputs of the two modules emphasises the increase of the printed character recognition level up to 89.1% for the M1 module and up to 89.8% for the M2 module, the number of errors decreasing vis-a-vis the RNA2 network response from 12.5% to 10.9%, and 10.2%, respectively. In order to set up the hidden layer of 90 neurons, a value of 92% was obtained at the output of the M2 analysis module.The performed model increased the printed character recognition rate by using the same primary information in a different manner. The validity and functionality of the suggested model are confirmed by experimental results.