The main problem of segmentation in spectral images that containing mixed pixels is addressed. Linear spectral unmixing is a procedure by which mixed pixels are decomposed into a collection of pure spectra, or endmembers, with their corresponding proportions, or abundances. Markov random field (MRF) is used to model the spatial correlation between pixels and segment the image into multiple classes. Pixels in each class have the same spectral values. A new numerical method was introduced to estimate the abundance and its parameters by using EM-algorithm and Gaussian mixture model which is termed as EM-MAP algorithm. A new solver, namely cross entropy (CE) was proposed for hyperspectral image unmixing. CE achieves higher performance of finding more global optima because of its stochastic property. The experiments show that CE can give more accurate segmentation results.

1. N. Keshava and J. Mustard, "Spectral unmixing," IEEE Signal Process. Mag. , vol. 19, no. 1, pp. 44–56, Jan. 2002.
2. N. Dobigeon, J. -Y. Tourneret, and C. -I. Chang, "Semi supervised linearspectral using a hierarchical Bayesian model for hyperspectral imagery,"IEEE Trans. Signal Process. , vol. 56, no. 7, pp. 2684–2696, Jul. 2008.
3. O. Eches, N. Dobigeon, C. Mailhes, and J. -Y. Tourneret, "Bayesian estimation of linear mixtures using the normal compositional model. Application to hyperspectral imagery," IEEE Trans. Image Process. ,vol. 19, no. 6, pp. 1403–1413, Jun. 2010.
4. J. Besag, "Spatial interaction and the statistical analysis of lattice systems," J. Royal Stat. Soc. Series B, vol. 36, no. 2, pp. 192–236, 1974.
5. F. Y. Wu, "The Potts model," Rev. Modern Phys. , vol. 54, no. 1, pp. 235–268, Jan. 1982.
6. N. Bali and A. Mohammad-Djafari, "Bayesian approach with hidden Markov modeling and mean field approximation for hyperspectral data analysis," IEEE Trans. Image Process. , vol. 17, no. 2, pp. 217–225, Feb. 2008.
7. O. Eches, N. Dobigeon, and J. -Y. Tourneret, "Enhancing hyperspectral image unmixing with spatial correlations," IEEE Trans. Geosci. Remote Sensing, vol. 49, no. 11, pp. 4239–4247, Nov. 2011.
8. L. Xu, M. I. Jordan, On Convergence Properties of the EM Algorithm for Gaussian Mixture, Neural Computation, 8, 1996, pp. 129-151.
9. M. A. T. Figueiredo, Bayesian Image Segmentation Using Gaussian Field Prior, EMMCVPR 2005, LNCS 3757, pp. 74-89.
10. R. Rubinstein, D. Kroese, eds. : The Cross-Entropy Method: A Unified Approach to Cominatorial Optimization, Monte-Carlo.

Hyperspectral Images Markov Random Field(mrf) Gaussian Mixture Model(gmm) Spectral Unmixing Cross Entropy(ce).