Reconhecimento de objetos bidimensionais usando modelagem poligonal.

Abstract

This dissertation presents a system capable of recognizing and locating objects in two-dimensional, isolated or overlapping, using the global and local attributes extracted from the polygonal models of the objects. The strategies used, by the system developed here, are closely linked to the way objects are modeled. A new algorithm for polygonal modeling of two-dimensional objects has been proposed. This algorithm has some relevant characteristics such as: it produces variable number of segments (sides of the polygon), depending only on the desired precision: it allows extraction of geometric measurements of the object model during the modeling process; integrates in a single step the processes of segmentation, modeling and extraction of characteristics. The operation of the algorithm is described in detail in this dissertation, experimental results are presented and conclusions are drawn regarding the advantages of the method and possible future applications.