中文摘要：

模拟人类视觉感知机制,提出了一种基于多尺度自回归滑动平均（MARMA,multiscale autoregressive and moving average model）模型和Markov随机场（MRF,markov random field）的合成孔径雷达（SAR）图像分割新方法。首先,分析人类视觉感知系统的工作机制和特点,利用SAR的成像机理,构建了SAR图像的金字塔结构和MARMA模型,以此模拟视觉过程中的空间尺度和朝向感知机制;然后,通过不同尺度上的MRF模型和改进的模拟退火（SA）算法实现更有效的多尺度分割策略。实验结果表明,本文提出的方法在SAR图像分割任务中有非常良好的表现。

英文摘要：

A new method of synthetic aperture radar （SAR） image segmentation is proposed, which simulates the mechanism of visual perception with multiscale auto-regressive and moving average （MARMA） model ,and Markov random field （MRF） models. Firstly,the characteristics and mechanisms of the visual perception system are analyzed. Secondly, the spatial scale and orientation perception mechanisms are simulated by constructing pyramid structure and MARMA model for SAR images. Then, the multiscale segmentation is achieved by the visual mechanism, the MRF model on different scales,and the improved simulated annealing algorithm. Experimental results on SAR images demonstrate that the proposed method has good performance for SAR image segmentation.