Computer Engineering and Applications ›› 2011, Vol. 47 ›› Issue (32): 211-214.

• 图形、图像、模式识别 • Previous Articles     Next Articles

Improvement to feature-level fusion of images based on MRF

NI Cui,GUAN Zequn,WANG Bin,ZHU Sujuan   

  1. Department of Surveying and Geo-informatics Engineering,Tongji University,Shanghai 200092,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2011-11-11 Published:2011-11-11

对马尔可夫随机场特征级图像融合的改进

倪 翠,关泽群,王 斌,朱素娟   

  1. 同济大学 测量与国土信息工程系,上海 200092

Abstract: This paper proposes a method of Markov random field to classify the remote sensing images based on the maximum a posteriori model.According the extraction of characteristic,image fusion course can be carried on.This method makes the mean and variance of the information in region of interest to be the probability parameters.Proper model is selected to get the exact solution of MAP rapidly in the course of iterated conditional mode.The whole fusion course can be finished.By means of two experiments,the proposed method shows its effectiveness and proves its superiority in remote sensing image fusion.

Key words: feature-level fusion, Maximum A Posteriori(MAP), Markov Random Field(MRF), Iterated Conditional Mode(ICM)

摘要: 提出了一种基于MAP的Markov随机场的图像融合方法。将感兴趣区特征的均值与方差作为马尔可夫随机场的概率参数,选取合适的模型,根据优化算法快速求得MAP解,完成图像初始标记过程,根据最大后验概率模型,对图像进行特征层融合。通过两组遥感图像的实验,证明MAP-MRF模型在遥感图像特征层融合中,具有较目前常用方法更好的效果。

关键词: 特征级图像融合, 最大后验概率, 马尔可夫随机场, 迭代条件模型