基于MRF场的SAR图像分割方法

提出了一种基于MRF（Markov Random field)模型的SAR（Synthetic Aperture Radar)图像分割算法，本算法利用ICM（（Iterative Conditional Mode)局部 优化方法，获得MAP（maximum a posteriori)准则下的图像分割结果。并引入了剔除外层数据的机制，用MSTAR（Moving and Stationary Target Acquisition and Recognition)数据进行实验，结果表明，算法能有效减少斑点噪声的影响将图像分割为目标，阴影，背景三部分，实验结果是令人满意的。

SAR Image Segmentation Based On Markov Random Field Model

In this paper, we propose a SAR image segmentation algorithm based on Markov Random Field Model. The model captures the image contextual information by making the assumption that the properties of a pixel are dependent upon nearby pixels and are independent of distant pixels. A two\|order neighborhood system is used in this paper. The segmentation result is obtained by applying ICM(Iterative Conditional Mode)algorithm. The ICM is an iterative algorithm. During the iteration sequence, each pixel is labeled according to the MAP(maximum a posteriori)criterion. After each pixel labeling step, the parameters of each segmentation class' probability density function are re-estimated from current set of pixel labels. The influence of speckle noise is further reduced by an outlier rejection method. Using this method, outlier pixels are excluded from current pixels' neighborhood. The algorithm is applied to SAR imagery from the MSTAR(Moving and Stationary Target Acquisition and Recognition)datasets. The result shows that this algorithm can efficiently reduce the influence of speckle noise, and segment the image into three parts of target, shadow and background. The experimental results are very satisfactory.