SciELO - Scientific Electronic Library Online

 
vol.6 issue3A New Algorithm for Computing Optical Flow and His Application to Image Registration author indexsubject indexsearch form
Home Pagealphabetic serial listing  

Services on Demand

Journal

Article

Indicators

Related links

  • Have no similar articlesSimilars in SciELO

Share


Computación y Sistemas

Print version ISSN 1405-5546

Comp. y Sist. vol.6 n.3 México Jan./Mar. 2003

 

Resumen de tesis doctoral

 

Methodologies for Reducing the Amount of Required Images Used for Articled–Object Recognition

 

Metodologías para la Reducción del Número de Imágenes Requeridas para el Reconocimiento de Objetos Articulados

 

Graduated: Luis Carlos Altamirano Robles
Instituto Mexicano del Petróleo, PIMAyC
Eje Central Lázaro Cárdenas 152, C.P. 07730
Del. Gustavo A. Madero, México, D. F.

E–mail: laltamir@imp.mx

Advisor 1: Leopoldo Altamirano
Instituto Nacional de Astrofísica, Óptica y Electrónica,
Puebla, Mexico

E–mail: robles@inaoep.mx

Advisor 2: Matías Alvarado Mentado
Instituto Mexicano del Petróleo
e–mail: matiasa@cic.ipn.mx

 

Graduated on March 19, 2002

Abstract

The appearance–based approaches are such that any object's model is made through a set of training images describing the object's appearance. In this PhD. thesis, the usage of non–uniform sampling is introduced for building this image set. Non–uniform sampling is held by a linear interpolation technique, which is used to determine the strictly necessary images. Main results are: a significant reduction in the quantity of necessary images for the object's model, as well as more precise models than those obtained by uniform sampling. Non–uniform sampling is used in conjunction with the eigenspaces technique for object recognition, producing a more efficient hybrid technique.

Keywords: non–uniform sampling, object recognition, appearance–based model, interpolation, eigenspaces.

 

Resumen

Los enfoques basados en apariencia construyen el modelo de un objeto, por medio de un conjunto de imágenes de entrenamiento que describe la apariencia del objeto. En esta tesis doctoral se propone el empleo del muestreo no–uniforme para generar tal conjunto de imágenes. El muestreo no–uniforme es soportado mediante una técnica de interpolación, que determina cuáles son las imágenes estrictamente requeridas para construir el modelo. Los resultados principales obtenidos con esta propuesta son: una reducción significativa de la cantidad de imágenes requeridas para construir el modelo del objeto, además de una mejora en la precisión de los modelos así generados, respecto a los obtenidos con muestreo uniforme. El muestreo no–uniforme es empleado junto con la técnica de espacios fundamentales (eigenspaces) para realizar el reconocimiento del objeto, obteniéndose una técnica híbrida más eficiente.

Palabras Clave: non–uniform sampling, object recognition, appearance–based model, interpolation, eigenspaces.

 

DESCARGAR ARTÍCULO EN FORMATO PDF

 

Acknowledgments

This research was sponsored in part by CONACyT under Ref. 131991 A. L.C. Altamirano was supported by CONACyT graduate scholarship No. 89635. The author would like to thank Dr. Matías Alvarado from Instituto Mexicano del Petróleo (PIMAyC), for reviewing and correcting a preliminary version of this work.

 

References

L.C. Altamirano–Robles. "Metodologías para la reducción del número de imágenes requeridas, para el reconocimiento de objetos articulados". Dissertation, PhD. Thesis, Centro de Investigación en Computación, Instituto Politécnico Nacional, México, D.F., March, 2002.        [ Links ]

L.C. Altamirano, L. Altamirano, and M. Alvarado. "Generation of N–parametric appearance–based models through non–uniform sampling". Submitted to Pattern Recognition Letters, The Netherlands, 2002.        [ Links ]

L.C. Altamirano, L. Altamirano, and M. Alvarado. "Non–Uniform Sampling For Improved Appearance–Based Models". Pattern Recognition Letters, Vol. 24, Issue 1–3, pp. 521–535, The Netherlands, January 2003.        [ Links ]

P.N. Belhumeur and D.J. Kriegman. "What is the Set of Images of an Object Under All Possible Ilumination Conditions?". International Journal of Computer Vision, Vol. no. 28, Issue No. 3, 245–260, 1998.        [ Links ]

T.F. Cootes, G.V. Wheeler, K.N. Walker and C.J. Taylor. "Coupled–View Active Appearance Models". The Eleventh British Machine Vision Conference, University of Bristol, September 2000.        [ Links ]

R. Epstein, P. Hallinan, and A. Yuille. "5±2 Eigenimages suffice: An empirical investigation of low–dimensional lighting models". In proceedings of IEEE Workshop on physics–based modeling in computer vision, 1995.        [ Links ]

J.J. Koenderink and A.J. Van Doorn. "The internal representation of solid shape with respect to vision". Biological Cybernetics, Vol. 32, 1979.        [ Links ]

B. Moghaddam and A. Pentland. "Probabilistic visual learning for object representation", in Early Visual Learning, Edited by Shree K. Nayar and Tomaso Poggio, New York Oxford, Oxford University Press, 1996.        [ Links ]

F. Mokhtarian and S. Abbasi. "Automatic Selection of Optimal Views in Multi–view Object Recognition". The Eleventh British Machine Vision Conference, University of Bristol, September 2000.        [ Links ]

H. Murase and S.K. Nayar. "Visual learning and recognition of 3–D objects from appearance". International Journal of Computer Vision, Vol. 14, No. 1, 5–24, January 1995.        [ Links ]

S.K. Nayar, H. Murase and S. Nene. "Parametric Appearance Representation". In Early Visual Learning, Edited by Shree K. Nayar and Tomaso Poggio, New York Oxford, Oxford University Press, 1996. Workshop, 1197–1205, 1997.        [ Links ]

R.C. Nelson and A. Selinger. "Experiments on (Intelligent) Brute Force Methods for Appearance–Based Object Recognition". DARPA Image Understanding.        [ Links ]

K. Ohba and K. Ikeuchi. "Recognition of the Multi Specularity Objects using the Eigen–Window". Proceedings of International Conference on Pattern Recongnition, August 1996.        [ Links ]

J. Pauli, M. Benkwitz and G. Sommer. "RBF Networks Appearance–Based Object Detection". Proceedings of ICANN, Paris, Volume 1, pp. 359–364, 1995.        [ Links ]

T. Poggio and D. Beymer. "Regularization Networks for Visual Learning". In Early Visual Learning, Edited by Shree K. Nayar and Tomaso Poggio, New York Oxford, Oxford University Press, 1996.        [ Links ]

A.  Pope and D. Lowe. "Learning Probabilistic Appearance Models for Object Recognition". In Early Visual Learning, Edited by Shree K. Nayar and Tomaso Poggio, New York Oxford, Oxford University Press, 1996.        [ Links ]

B. Schiele and J.L. Crowley. "Recognition without Correspondence using Multidimensional Receptive Field Histograms". International Journal of Computer Vision 36(1), 31–50, 2000.        [ Links ]

L.G. Shapiro and M.S. Costa. "Appearance–Based 3D Object Recognition". Object Representation in Computer Vision, Vol. I, in Proceedings of International NSF–ARPA Workshop, New York City, NY, USA, Dec. 1994. 51–64.        [ Links ]

M.R. Stevens and J.R. Beveridge. "Integrating graphics and vision for object recognition". Kluwer Academic Publishers, 2000.        [ Links ]

M. Turk and A. Pentland. "Eigenfaces for Recognition". J. Cognitive Neuroscience, Vol. 3, no. 1, 71–86, 1991.        [ Links ]

Creative Commons License All the contents of this journal, except where otherwise noted, is licensed under a Creative Commons Attribution License