去除SAR图像斑点的改进型自适应各向异性扩散滤波方法

An enhanced adaptive SAR speckle filter based on anisotropic diffusion

针对目前基于各向异性扩散技术的 SAR 图像斑点滤波方法的扩散系数与迭代过程控制不够精确的缺点,提出了一种改进的自适应各向异性扩散滤波方法,该方法结合有效视数(ENL)和梯度两种参数,使扩散系数能够自适应不同地物与细微结构,并通过准确地获取迭代过程中平坦区域有效视数以及扩展扩散的临近点个数,更精确地控制迭代过程。实验表明,这种改进的方法在去除斑点噪声、保持点目标和细节边缘信息等方面均优于传统的增强 Lee 滤波法、PM 滤波法和 SRAD 各向异性扩散法。

Considering that the present anisotropic diffusion-based SAR speckle filters＇ diffusion coefficients and the control of iteration process are not accurate, the paper proposes an improved adaptive anisotropic diffusion filter. It utilizes the equivalent number of looks （ENL） and gradient parameters together to adapt the diffusion coefficients to different classes of SAR images. Furthermore, it calculates exactly the ENL value of the homogeneous region in each iteration and expands the number of neighbor pixels to control the process of iteration more accurately. As Compared with other anisotropic diffusion speckle reduction methods such as the enhanced Lee filter, the PM filter and the speckle reduction anisotropic diffusion （SRAD） filter, the proposed method can reduce SAR specrkles and preserve the information of detailed edges and the point targets more effectively.