利用马尔柯夫随机场图模型的变化像斑类别判定方法
Determination of New Class Properties of the Changed Image Segments Using MRF Graph Model
-
摘要: 提出了一种利用马尔柯夫随机场(MRF)图模型获取变化像斑变化后类别的方法。用吉布斯(Gibbs)分布描述MRF图模型,通过计算最大后验概率获取变化像斑类别,以此类别为初始值,通过建立像斑类别空间关系概率矩阵,在初始结果基础上进一步判定变化像斑的类别。实验结果证实了该方法的可行性。Abstract: We give a proposal to obtain the new class property of each changed image segment using MRF graph model.The new class property of each changed image segment can be gotten by calculating maximum posterior probability through MRF graph model descripted by Gibbs distribution.The result could be regarded as the initial result and then be further analyzed using class spatial-relationship probability matrix.The experimental result shows the feasibility of this method.