ASM Modeling for Training System and Handwritten Recognition

In the area of Pattern Recognition, specifically in the field of Handwriting Recognition, will be presented modeling a system of training and recognition of handwritten characters and figures by means an Abstract State Machine (ASM). The processing operation is to perform the thinning in the singular regions of the track pattern handwritten character after determining the thinning in the regular regions. In fact, it is considered a character of the Latin alphabet, the letter "R" in capital letters, or the digit "4", chosen because they contain all the singularities that are possible to find the other characters or figures, and applies them to a technique structural decomposition intended for training and recognition using Hidden Markov Model (HMM). Of course, the same character written by different authors, or by the same author at different times, produces a different result. So, the training activities are more properly configured with an operation for classifying handwritten characters more or less similar, based on slight differences in slope of the stroke. The strokes are the parts of the track, the character components manuscript. The classification of characters allows obtaining a knowledge base to be used in the recognition phase.