Un enfoque bayesiano para la extracción de características y agrupamiento en visión artificial

Abstract

This thesis focuses on feature extraction and perceptual grouping in computer vision.Our main purpose has been to formulate and test new methods for feature extraction (mainly those related to corner identification and junction classification) and grouping (through junction connection).The context of application of these methods, robot vision, imposes special constrains over their applicability and thus, the following requirements are observed: efficiency, robustness, and flexibility.These requirements are present in almost all the techniques presented in this thesis: Junction classification is performed by greedy method that relies on sound statistics; Junction grouping is performed by a path-searching method of linear complexity on average, which is based on pruning conditions to constrain the search of stable paths, and such conditions rely on statistical information; and junction grouping yields speeding up the voting scheme used to find the relative orientation between the robot and the environment.However, the range of application of the proposed method is not restricted to infer relative orientation, and can be extended to other tasks like image segmentation, depth estimation or object recognition.This thesis is structured in three parts which cover junction classification, grouping and relative orientation, yielding a bottom-up exposition: ndice General Introduccin1.1 Visin orientada a tareas frente a visin reconstructiva . . . . . . . . .