【摘要】 在多尺度Markov模型的基础上,提出了一种新的用于SAR图像无监督分割的上下文融合分割方法。该方法充分考虑了SAR图像分布的统计特性,用基于混合Rayleigh分布的多尺度Markov模型对待分割图像建模,并直接根据待分割图像用迭代条件估计算法来训练模型的参数。然后以上下文向量的形式提出了四种不同的上下文模型,并用这四种上下文模型分别对待分割图像的多尺度图像信息进行自上而下的融合,最终得到四种不同的分割结果。实验表明,该方法进一步提高了SAR图像分割结果的精度。更多还原

【Abstract】 A new unsupervised context-based fused segmentation algorithm for SAR imagery was proposed based on Multiscale Markov model.The method fully considered statistical characterization of SAR imagery and approximated segmented imagery with mixture rayleigh distribution.The model parameters could be straightly trained by iterative conditional estimation algorithm based on segmented image.Then we proposed four different context models to fuse multiscale image information.Finally,four different segmentation separately were obtained.Simulations on synthetic image and SAR imagery indicate that the new approach improves segmentation accuracy.更多还原