摘要
MRF图像分割算法由于其良好的抗噪性,在图像分割邻域应用广泛,但其存在过度分割现象,导致边缘分割模糊定位不准确。针对这一问题,文章首先在传统的MRF图像分割算法中引入可变权重的参数来连接标记场模型与特征场模型,使得两种模型之间形成一种平衡,获取可保持图像边缘、图像重要细节和具有区域一致性的分割结果。然后在边缘处自适应地引入边缘惩罚函数,调整势函数的能量对能量函数的贡献,减少分割时对边缘的模糊,提高对边缘的定位精度。通过实验结果分析表明,所提出的优化的MRF影像分割算法比传统的ICM迭代计算MRF分割算法和变权重的MRF分割算法具有更高分割精度。
Markov Random Field(MRF)algorithm is widely used in image segmentation because of its excellent noise immunity.However,its image over-segmentation phenomenon leads to edge fuzzy segmentation result and inaccurate positioning accuracy.To solve this problem,the paper introduced adaptive weight parameters connecting the mark field model and feature field model.The adoptive weight parameters form a balance of state between the two models,and then the maintained image edge and more important details segmentation results as well as the consistency of regional segmentation results are obtained.In addition,an edge penalties function was introduced to deal with the problem of edge fuzzy and inaccurate positioning accuracy,through adaptive adjusting the energy of clique potential of clique to adjust the energy function,reducing edge fuzzy results and improving positioning accuracy of the edge.The experimental result showed that the proposed algorithm has higher segmentation accuracy than the traditional ICM and the adaptive weight parameters MRF image segmentation algorithm.
作者
郭建华
杨帆
谭海
王竞雪
Jianhua Guo;Fan Yang;Hai Tan;Jingxue Wang(School of Geomatics,Liaoning Technical University,Fuxin,Liaoning 123000,China;Satellite Surveying and Mapping Application Center,National Administration of Surveying,Mapping and Geoinformation,Beijing 100048,China)
关键词
影像分割
马尔科夫随机场
变权重
边缘惩罚函数
Image Segmentation
Markov Random Field
Adaptive Weight Parameter
Edge Penalties Function
