Advanced search
1 file | 4.72 MB
Author
Organization
Abstract
In this chapter we present a family of techniques based on the principle of the Local Binary Pattern (LBP) technique. This family is called the Geometric Local Textural Patterns (GLTP). Classical LBP techniques are based on exploring intensity changes around each pixel in an image using close neighbourhoods. The main novelty of the GLTP techniques is that they explores intensity changes on oriented neighbourhoods instead of on close neighbourhoods. An oriented neighbourhood describes a particular geometry composed of points on circles with different radii around the center pixel. A digital representation of the points on the oriented neighbourhood defines a GLTP-code. Symmetric versions of the geometries around the pixel are assessed the same GLTP code. Each pixel in the image is assigned a set of GLTP-codes, each for a particular geometry. The texture of an image is characterized with a GLTP histogram of the occurrences of the GLTP-codes on the whole image. We explain the principle of the techniques using the simplest case, called the Geometric Local Binary (GLBP) technique, which is based on boolean comparisons. Then we present variations of this technique to enlarge the family of GLTP techniques. We quantify the texture difference between a pair or images or regions by computing the divergence between their corresponding GLTP-histograms using an adaptation of the Jensen-Shannon entropy.
Keywords
image processing, LBP, Texture

Downloads

  • (...).pdf
    • full text
    • |
    • UGent only
    • |
    • PDF
    • |
    • 4.72 MB

Citation

Please use this url to cite or link to this publication:

Chicago
Orjuela Vargas, Sergio Alejandro, Juan Pablo Yañez, and Wilfried Philips. 2014. “The Geometric Local Textural Patterns (GLTP) Technique.” In Local Binary Patterns : New Variants and New Applications, ed. Sheryl brahnam, Lakhmi C. Jain, Loris Nanni, and Alessandra Lumini, 506:30–70. 1st ed. Heidelberg, Germany: Springer.
APA
Orjuela Vargas, S. A., Yañez, J. P., & Philips, W. (2014). The Geometric Local Textural Patterns (GLTP) technique. In S. brahnam, L. C. Jain, L. Nanni, & A. Lumini (Eds.), Local binary patterns : new variants and new applications (1st ed., Vol. 506, pp. 30–70). Heidelberg, Germany: Springer.
Vancouver
1.
Orjuela Vargas SA, Yañez JP, Philips W. The Geometric Local Textural Patterns (GLTP) technique. In: brahnam S, Jain LC, Nanni L, Lumini A, editors. Local binary patterns : new variants and new applications. 1st ed. Heidelberg, Germany: Springer; 2014. p. 30–70.
MLA
Orjuela Vargas, Sergio Alejandro, Juan Pablo Yañez, and Wilfried Philips. “The Geometric Local Textural Patterns (GLTP) Technique.” Local Binary Patterns : New Variants and New Applications. 1st ed. Ed. Sheryl brahnam et al. Vol. 506. Heidelberg, Germany: Springer, 2014. 30–70. Print.
@incollection{3225212,
  abstract     = {In this chapter we present a family of techniques based on the principle of the Local Binary Pattern (LBP) technique. This family is called the Geometric Local Textural Patterns (GLTP). Classical LBP techniques are based on exploring intensity changes around each pixel in an image using close neighbourhoods. The main novelty of the GLTP techniques is that they explores intensity changes on oriented neighbourhoods instead of on close neighbourhoods. An oriented neighbourhood describes a particular geometry composed of points on circles with different radii around the center pixel. A digital representation of the points on the oriented neighbourhood defines a GLTP-code. Symmetric versions of the geometries around the pixel are assessed the same GLTP code. Each pixel in the image is assigned a set of GLTP-codes, each for a particular geometry. The texture of an image is characterized with a GLTP histogram of the occurrences of the GLTP-codes on the whole image. We explain the principle of the techniques using the simplest case, called the Geometric Local Binary (GLBP) technique, which is based on boolean comparisons. Then we present variations of this technique to enlarge the family of GLTP techniques. We quantify the texture difference between a pair or images or regions by computing the divergence between their corresponding GLTP-histograms using an adaptation of the Jensen-Shannon entropy.},
  author       = {Orjuela Vargas, Sergio Alejandro and Ya{\~n}ez, Juan Pablo and Philips, Wilfried},
  booktitle    = {Local binary patterns : new variants and new applications},
  editor       = {brahnam, Sheryl and Jain, Lakhmi C. and Nanni, Loris  and Lumini, Alessandra},
  isbn         = {9783642392887},
  issn         = {1860-949X},
  keyword      = {image processing,LBP,Texture},
  language     = {eng},
  pages        = {30--70},
  publisher    = {Springer},
  series       = {Studies in Computational Intelligence},
  title        = {The Geometric Local Textural Patterns (GLTP) technique},
  url          = {http://www.springer.com/engineering/computational+intelligence+and+complexity/book/978-3-642-39288-7},
  volume       = {506},
  year         = {2014},
}