基于主动轮廓模型的合成孔径雷达图像分割方法

针对合成孔径雷达(synthetic aperture radar,SAR)图像中的乘性相干斑噪声影响分割效果的问题,提出了一种基于主动轮廓模型的分割方法。该方法将改进的细节保持各向异性扩散(IDPAD)滤波和SBGFRLS模型相融合,构造了一个边迭代演化边抑噪的改进的演化微分方程。在每一次迭代演化中,先借助改进的演化微分方程演化水平集函数,然后利用高斯滤波器正则化水平集函数,最后通过检查水平集函数的收敛性判定分割是否完成。实验结果表明,与经典的分割方法相比,本文方法在保护边缘的同时减少了乘性相干斑噪声对SAR图像分割的影响,减少了误检轮廓,且对初始轮廓不敏感。

SAR Image Segmentation Method Based on Active Contour Model

Aiming at the problem that multiplicative speckle noise affects the segmentation effect in synthetic aperture radar (SAR) images, a new segmentation method based on improved active contour model is proposed. The method combines the improved detail-preserving anisotropic diffusion (IDPAD) filtering with the SBGFRLS model to construct an improved evolutionary differential equation with iterative evolution and noise suppression. In each iteration, the level set function is first evolved by means of the improved evolutionary differential equation, then the Gaussian filter is used to regularize the level set function, and finally the convergence of the level set function is checked to determine whether the segmentation is complete. The experimental results and comparisons with the classical segmentation method show that the proposed method reduces the influence of multiplicative speckle noiseon SAR image segmentation,reduces false detection profile and is insensitive to the initial contour.