论文部分内容阅读
基于Markov随机场(MRF,Markov Random Field)的SAR图像分割方法利用了SAR图像的灰度和结构信息,能在分割过程中有效抑制斑点噪声,获得较高的分割精度.但这类方法的缺点是模拟退火的计算量很大.针对该问题,提出了一种基于快速退火MRF的SAR图像分割处理方法.该方法根据SAR图像Gibbs分布的特性,在求取全局最优解时,首先寻找邻域系统中占有支配地位的某种标记,若存在占支配地位的标记,用此标记更新状态;反之,则沿用传统模拟退火的方法随机更新状态.由于该方法引入基于Gibbs分布的先验判决进行系统状态更新,因此能够快速求得全局最优解.最后对真实SAR图像进行处理,处理结果验证了算法的有效性.
The SAR image segmentation method based on Markov random field (MRF) makes use of the gray level and structure information of SAR images and effectively restrain the speckle noise and obtain higher segmentation accuracy in the segmentation process. However, the shortcomings of such methods Which is a large computational burden of simulated annealing.Aiming at this problem, a fast SAR image segmentation processing method based on fast annealing MRF is proposed.According to the characteristics of Gibbs distribution of SAR image, when finding the global optimal solution, If there is dominant mark, the mark is used to update the state, otherwise, the state is randomly updated according to the traditional simulated annealing method.Because the method introduces a priori judgment based on Gibbs distribution The state of the system is updated, so the global optimal solution can be obtained quickly.Finally, the real SAR image is processed, and the processing results verify the effectiveness of the algorithm.