基于MRF高斯混合模型的海浪纹理背景目标区域分割

Object region segmentation of sea wave texture background based on MRF Gaussian-mixed model

为进行海洋图像处理,根据海面背景的特点,提出1种基于马尔科夫随机场(MakovRan-domField,MRF)高斯混合模型的海浪纹理背景目标区域分割方法.通过对海平面特征的分析,建立MRF先验模型;运用MRF和Gibbs分布理论,建立海浪纹理的MRF高斯混合模型;用期望极大化方法获取海浪纹理的参数;根据最小后验能量准则实现海浪纹理背景目标区域的分割.与采用经典分割方法进行对比表明,该方法可行.

In order to process the ocean image, a method of object region segmentation against the sea wave texture background based on Markov Random Field （MRF） and Gaussian-mixed model is proposed according to the characteristics of the sea wave background. A prior MRF model is established by analyzing the characteristics of the sea wave background; An MRF Gaussian-mixed model about the sea wave texture is put forward based on MRF and Gibbs Distributing theory; The sea wave texture parameters are obtained by the means of Expectation Maximization algorithm; The object region segmentation against sea texture background is realized by the rule of Minimum Posterior Energy. Compared with the classical segmentation method, the method shows its feasibility.