Um sistema de visão inteligente para detecção e reconhecimento de peças em um tabuleiro de xadrez em tempo real.

Abstract

This dissertation presents the development of a intelligent vision system, based on artificial neural networks, developed to detect and to recognize pieces of a chessboard in real time. A software, denominated NeuroMorfo, was implemented to accomplish several tasks as acquisition of images in real time, image processing, detection on-line and off-line of chess pieces, and off-line recognition of these pieces. Regarding the detection system, modifications were made in the original system, resulting in a new detection algorithm with concepts of the Intelligent Forecast System (SPI). Regarding the recognition system, studies were accomplished related to the necessary stages to the process of pattern recognition that facilitated its development. The performance of the detection and recognition systems is evaluated starting from a series of accomplished tests.