Maximum Entropy Models for Skin Detection

We consider a sequence of three models for skin detection built from a large collection of labelled images. Each model is a maximum entropy model with respect to constraints concerning marginal distributions. Our models are nested. The first model, called the baseline model is well known from practitioners. Pixels are considered independent. Performance, measured by the ROC curve on the Compaq Database is impressive for such a simple model. However, single image examination reveals very irregular results. The second model is a Hidden Markov Model which includes constraints that force smoothness of the solution. The ROC curve obtained shows better performance than the baseline model. Finally, color gradient is included. Thanks to Bethe tree approximation, we obtain a simple analytical expression for the coefficients of the associated maximum entropy model. Performance, compared with previous model is once more improved.

[1]  Peter C. Doerschuk,et al.  Tree Approximations to Markov Random Fields , 1995, IEEE Trans. Pattern Anal. Mach. Intell..

[2]  J. Besag.  On the Statistical Analysis of Dirty Pictures , 1986 .

[3]  James M. Rehg,et al.  Statistical Color Models with Application to Skin Detection , 1999, Proceedings. 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No PR00149).

[4]  Song-Chun Zhu,et al.  Equivalence of Julesz Ensembles and FRAME Models , 2000, International Journal of Computer Vision.

[5]  L. Younes.  Estimation and annealing for Gibbsian fields , 1988 .

[6]  Jun Zhang.  The mean field theory in EM procedures for Markov random fields , 1992, IEEE Trans. Signal Process..

[7]  Alfredo Gardel Vicente,et al.  Unsupervised and adaptive Gaussian skin-color model , 2000, Image Vis. Comput..

[8]  James Ze Wang,et al.  Classifying Objectionable Websites Based on Image Content , 1998, IDMS.

[9]  Leszek Wojnar,et al.  Image Analysis , 1998 .

[10]  Shigeru Akamatsu,et al.  Automatic detection of human faces in natural scene images by use of a skin color model and of invariant moments , 1998, Proceedings Third IEEE International Conference on Automatic Face and Gesture Recognition.

[11]  Gerhard Winkler,et al.  Image analysis, random fields and dynamic Monte Carlo methods: a mathematical introduction , 1995, Applications of mathematics.

[12]  Gilles Celeux,et al.  EM procedures using mean field-like approximations for Markov model-based image segmentation , 2003, Pattern Recognit..

[13]  Donald Geman,et al.  Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images , 1984, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[14]  Fabio Divino,et al.  Penalized Pseudolikelihood Inference in Spatial Interaction Models with Covariates , 2000 .

[15]  Song-Chun Zhu,et al.  Filters, Random Fields and Maximum Entropy (FRAME): Towards a Unified Theory for Texture Modeling , 1998, International Journal of Computer Vision.

[16]  E. Jaynes.  Information Theory and Statistical Mechanics , 1957 .

[17]  Anders Martin-Löf,et al.  The equivalence of ensembles and the Gibbs phase rule for classical lattice systems , 1979 .

[18]  Thomas M. Cover,et al.  Elements of Information Theory , 2005 .

[19]  Judea Pearl,et al.  Probabilistic reasoning in intelligent systems - networks of plausible inference , 1991, Morgan Kaufmann series in representation and reasoning.

[20]  T. Stephenson.  Image analysis , 1992, Nature.

[21]  James Ze Wang,et al.  System for screening objectionable images , 1998, Comput. Commun..

[22]  Shigeru Akamatsu,et al.  Comparative performance of different skin chrominance models and chrominance spaces for the automatic detection of human faces in color images , 2000, Proceedings Fourth IEEE International Conference on Automatic Face and Gesture Recognition (Cat. No. PR00580).