摘要
SAR图像极低的信噪比以及乘性噪声给SAR图像的边缘检测带来了较大的困难。提出了一种针对SAR图像边缘的自适应贝叶斯检测方法。该方法利用广义高斯马尔可夫随机场作为局部均值的先验概率分布模型,利用贝叶斯准则推导了局部均值的最大后验概率估计。广义高斯马尔可夫随机场模型参数估计和局部均值估计采用联合迭代技术进行求解。边缘检测器的参数采用接收机操作性能曲线和卡方检验进行选择。基于实测SAR数据的仿真实验结果表明,本文的边缘检测算子是有效的,并优于已有的SAR图像边缘检测算子。
Due to the low signal-to-noise ratios and the muhiplicative nature of speckle, edge detection is particularly difficult for synthetic aperture radar （SAR） imagery. A new spatially adaptive Bayesian method based on Markov random field （MRF） is proposed to detect edges in SAR imagery. The generalized Gaussian MRF （GGMRF） is employed as a prior distribution to develop a maximum a posteriori probability estimator for the local mean power. A method of jointly and iteratively estimating the local mean power and the hyperparameters of GGMRF model is introduced. The receiver operating characteristics analysis and Chi-square test are utilized to determine optimal parameters of the edge detector. Experiments are carried out by using real SAR images, and the results show that the proposed edge detector is effective and performs favorably in comparison with the existing popular edge detectors in most cases.
出处
《宇航学报》
EI
CAS
CSCD
北大核心
2012年第12期1832-1839,共8页
Journal of Astronautics
基金
国家自然科学基金(61002045)
关键词
最大后验概率估计
广义高斯马尔可夫随机场
边缘检测
合成孔径雷达
比率
Maximum a posteriori probability estimation
Generalized Gaussian-Markov random field
Edgedetection
Synthetic aperture radar
Ratio
