An image segmentation algorithm of adaptive prior MRF models

被引：0
|
作者：
机构：
[1] Information and Control Engineering School, Xi'an University of Architecture and Technology
[2] School of Electronics and Information Engineering, Xi'an Jiaotong University
来源：
Ren, Q. | 1600年 / Xi'an Jiaotong University卷 / 47期
关键词：
Adaptive prior; Belief propagation algorithm; Image segmentation; Markov random field;
D O I：
10.7652/xjtuxb201310011
中图分类号：
学科分类号：
摘要：
A segmentation algorithm based on a local adaptive prior Markov random field (MRF) model is proposed to solve the problem that the global homogeneous prior MRF model is inefficient to utilize the local statistic feature of nature images for image segmentation. The algorithm is based on Bayesian theory, utilizes local prior Potts model to represent image local features, and builds a local adaptive prior MRF model. A modified local region belief propagation (BP) algorithm is proposed over the MRF model, hence, local region features of an image are spreaded in global. The segment results of an image are obtained using a maximum posteriori (MAP) criterion. Experiment results show that the proposed adaptive prior MRF model generates more accurate segment results on the noise and texture of images than the global homogeneous prior MRF model does, and the proposed model has strong robustness on the interference of image noise or texture. The segmentation algorithm generates more accurate segmentation with less iterations and a short time.
引用
收藏
页码：62 / 67
页数：5
相关论文
共 16 条
  • [1] Gu D., Sun J., EM image segmentation algorithm based on an inhomogeneous hidden MRF model, Vision Image and Signal Processing, 152, 2, pp. 184-190, (2005)
  • [2] Alexandre L.M.L., Nelson D.A.M., Alberto T., Et al., Spatially non-homogeneous Potts model parameter estimation on higher-order neighborhood systems by maximum pseudo-likelihood, Proceedings of the 23rd Annual ACM Symposium on Applied Computing, pp. 1733-1737, (2008)
  • [3] Zhang X., Lam K.-M., Shen L., Image magnification based on a blockwise adaptive Markov random field model, Image and Vision Computing, 26, 9, pp. 1277-1284, (2008)
  • [4] Zhu F., Luo L., Song Y., Et al., Adaptive spatially neighborhood information Gaussian mixture model for image segmentation, Journal of Computer Research and Development, 4, 11, pp. 2000-2007, (2011)
  • [5] Besag J., On the statistical analysis of dirty picture, Journal of the Royal Statistical Society, 48, 3, pp. 259-302, (1986)
  • [6] Geman S., Geman D., Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images, IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-6, 6, pp. 721-741, (1984)
  • [7] Marroquin J.L., Probabilistic solution of inverse problems, (1985)
  • [8] Joshi M.V., Chaudhuri S., Panuganti R., A learning-based method for image super-resolution from zoomed observation, IEEE Transactions on System, Man, and Cybernetics, 35, 3, pp. 527-537, (2005)
  • [9] Peng B., Zhang L., Zhang D., Et al., Image segmentation by iterated region merging with localized graph cuts, Pattern Recognition, 44, 10-11, pp. 2527-2538, (2011)
  • [10] Xu L., Jordan M.I., On convergence properties of the EM algorithm for Gaussian mixtures, Neural Computation, 8, 1, pp. 129-151, (1996)
12