Algoritmos clássicos de aprendizado de maquina aplicados ao problema do reconhecimento de imagens

Abstract

The interest for the development of applications involving computer vision can be observed by the increasing number of systems that seek to explore the benefits joined by this branch of computer science. Example of that is the use of this type of system in equipments of daily use, which often pass in an unnoticed way for the people, as in some models of digital cameras and automobiles. Computer vision systems involve several stages in its process, from image acquisition in a digital format to the decision-making, which looks for, through the integration of techniques of images processing and artificial intelligence, to supply the computer systems means so that they are capable to interact through the visual perception with the environment around them. The focus of this work is in the classification and recognition stage of a computer vision system, where the objective is to classify the different elements that compose a certain scene based on extracted information from its digital representation. Here different forms to accomplish this stage will be approached, using some of the techniques of artificial intelligence, specifically machine learning, and analyzing the behavior of each when applied to the group of selected images.