Sistema para tomada de decisão baseado na geração de imagens estereoscópicas e reconhecimento de padrões circulares

Abstract

In this work, it is presented a system for decision-making based on stereoscopic image generation and recognition of circular patterns. For stereoscopic imaging are used two low cost Web cam in canonical configuration and block correspondence, which is used to find out the correspondence between images from a scene of interest. Also, the fast Hough transform is used to recognize circular patterns, which are identified and corrected based on their diameters with depth information obtained from both stereoscopic imaging and a calibration factor. The results obtained with simulated patterns and with composed real scenes with circular patterns in varied situations of special distribution illustrate the method potentiality for decision-making in procedures, which involve circular pattern recognition.