摘要
讨论了基于马尔可夫随机场的遥感图像分割．根据卫星遥感图像的特点，建立了相应的基于马尔可夫随机场的图像分割模型．由此将图像分割问题转化成图像标记问题， 并进一步转化成求图像的最大后验概率估计的问题． 本文引进了一种基于博弈理论的决定性退火算法， 可以用该算法对图像进行标记， 该算法收敛于局部最大， 在实验中取得了很好的效果．
The segmentation of Remote Sensing Image based Markov Random Field (MRF) is discussed in this article. According to the characteristics of satellite image, a segmentation model based on MRF is established. Thus, the problem of segmentation is converted to the problem of image symbolizing, and finally converted to the solution of Maximum A Posterior(MAP). In this paper we provide a deterministic algorithm which is based on 'Game Theory', use it to symbolize the image and prove that it can converge to a local optimum. Good result is got in the experiment.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
1999年第B11期11-15,共5页
Journal of Southeast University：Natural Science Edition
基金
国防科工委重点项目
关键词
马氏随机场
遥感图象
图像分割
模拟退火
Markov random field
image segmentation
simulated annealing
maximum a posterior
game theory
