This project is a study about the practical application techniques for the development of an Automatic Passport Identification Number Recognition System. The system proposed in this work is based on an image classifier developed by supervised learning techniques with convolutional neuronal networks. The basic image processing techniques are studied in this project due to its important role in the character recognition problems. In this document, it is proposed a system based on a simple convolutional neural network capable of executing the text recognition and successively the segmentation, analysis and classification of the characters; thus registering the number that appears in the image. For the development of this system it has been used the Matlab software and the Neural Network and Image Processing Toolboxes.