Publication: Modelado automático tridimensional del entorno y extracción de características
Loading...
Identifiers
Publication date
2013
Defense date
2013-03-11
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
La visión 3D ha sido uno de los campos que más se han visto extendidos en los últimos años. Los últimos avances tecnológicos han propiciado una mayor facilidad
y fiabilidad a la hora de obtener la información del entorno.
Hasta ahora, problemas como detectar el suelo o el entorno requerían de un hardware muy costoso, además de aparecer numerosas dificultades como errores de
sincronización entre el hardware y la unidad de proceso. Actualmente, dichos problemas
se encuentran paliados, en parte, gracias a la aparición de cámaras de bajo
coste, como por ejemplo el dispositivo Microsoft Kinect, que poseen la capacidad
necesaria para desempeñar esas funciones. Por otro lado, han aparecido librerías
de desarrollo enfocadas especialmente para el campo de la visión 3D.
En este proyecto se propone un algoritmo capaz de procesar la información recibida
de una cámara Kinect con el fin de detectar suelos lisos y transitables.
Además, el algoritmo es capaz de superar las limitaciones intrínsecas de Kinect,
aumentando el rango de funcionamiento mediante técnicas de aprendizaje. Para
ello se realizarán pasos intermedios, que incluyen el análisis, aprendizaje e interpretación de los parámetros de color. Para esta tarea se utilizar a tanto el lenguaje
de programación C++ como librerías externas tales como OpenCV o PCL, además del apoyo de la herramienta matemática Matlab.
Artificial 3D Vision is one of the research fields which has been extended the most in recent years thanks to current technological advances that have led to a greater ease and reliability when obtaining information from the environment. Up to now, detecting problems like soil or environment required a hardware very expensive, besides appearing numerous problems as synchronization errors between the hardware and the processing unit. Currently, these problems are alleviated in part by the emergence of low-cost cameras, such as Microsoft Kinect, which have the capacity to perform these functions. On the other hand, have appeared specially focused development libraries for 3D computer vision. In this project we propose an algorithm capable of processing the information received from a Kinect camera to detect smooth and walkable zones. We also implement an algorithm able to remove the intrinsic limitations of Kinect, increasing the maximum operating distance. This will involve intermediate steps, including the analysis, learning and interpretation algorithms of color parameters. For this task we use both the C++ programming language and external libraries such as OpenCV or PCL, and support from the Matlab mathematical tool.
Artificial 3D Vision is one of the research fields which has been extended the most in recent years thanks to current technological advances that have led to a greater ease and reliability when obtaining information from the environment. Up to now, detecting problems like soil or environment required a hardware very expensive, besides appearing numerous problems as synchronization errors between the hardware and the processing unit. Currently, these problems are alleviated in part by the emergence of low-cost cameras, such as Microsoft Kinect, which have the capacity to perform these functions. On the other hand, have appeared specially focused development libraries for 3D computer vision. In this project we propose an algorithm capable of processing the information received from a Kinect camera to detect smooth and walkable zones. We also implement an algorithm able to remove the intrinsic limitations of Kinect, increasing the maximum operating distance. This will involve intermediate steps, including the analysis, learning and interpretation algorithms of color parameters. For this task we use both the C++ programming language and external libraries such as OpenCV or PCL, and support from the Matlab mathematical tool.
Description
Keywords
Modelado automático, Visión 3D, Proceso de imágenes, Algoritmos, Robots autónomos móviles