Generalized Gaussian density for skin detection in DCT domai

  • Authors:
  • Sanaa Ghouzali;Sheila Hemami;Mohammed Rziza;Driss Aboutajdine

  • Affiliations:
  • School of Electrical and Computer Engineering, Cornell University, Ithaca, New York and GSCM, Faculté des Sciences, Université Mohammed V-Agdal, Rabat, Morocco;School of Electrical and Computer Engineering, Cornell University, Ithaca, New York;GSCM, Faculté des Sciences, Université Mohammed V-Agdal, Rabat, Morocco;GSCM, Faculté des Sciences, Université Mohammed V-Agdal, Rabat, Morocco

  • Venue:
  • Machine Graphics & Vision International Journal
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we propose a highly efficient algorithm to model the human skin color. The algorithm involves generating a discrete Cosine transform (DCT) at each pixel location, using the surrounding points. The DCT coefficients incorporate the pixel color and texture information to distinguish between skin and non-skin. A generalized Gaussian distribution (GGD) is used in this framework to model the DCT coefficients at low frequencies. Next, the model parameters are estimated using the maximum-likelihood (ML) criterion applied to a set of training skin samples. Finally, each pixel is classified as skin if its likelihood ratio exceeds some threshold. The experimental results show that our model avoids excessive false detection while still retaining a high degree of correct detection.