SAR图像分割的G_A^0统计模型-水平集方法

SAR Image Segmentation Based on G_A^0 Statistical Model-level Set method

由于合成孔径雷达(Synthetic Aperture Radar,SAR)图像易受相干斑噪声的影响,光学图像的分割方法并不适用于SAR图像,更不能获得精确的分割结果,首先基于G0A统计模型定义能量映射函数以代替像素值进行后续的处理,该方法可以减小相干斑的影响;其次,使用水平集算法对处理后的图像进行分割处理,选用了一种形式更为简单的水平集函数,并可以较容易地推广到多区域SAR图像分割情况.实验结果表明该方法可以减少相干斑噪声对SAR图像分割过程的不良影响,具有较好的准确性.

To solve the problem that optical image segmentation method is not suitable for SAR image and cannot obtain the accurate segmentation results ,which is affected by synthetic aperture radar image′speckle noise ,a SAR image segmentation method based on G0A statistical model and level set is presented in this paper .Firstly ,a energy mapping function based on G0A statistical model is defined to instead of pixel values for subsequent image processing , which can reduce the speckle noise′influence;Secondly ,using level set method to finish images′segmentation ,this article uses a more simple energy function which is easy to be used to deal with multiregion SAR images′segmentation .Experimental results show that this method can reduce speckle noise′ influence in SAR image segmentation process ,and has better accuracy and adaptive .