抑制SAR图像相干斑的自适应红黑窗滤波算法

Adaptive red-black window algorithm for SAR image speckle reduction

针对经典空域滤波算法处理SAR图像时,在同一滑动窗内完全按同质区域性质处理数据而导致图像细节信息损失的情况,提出一种新的滑动窗结构的算法。该算法选取与窗口中心像素统计特性相近的像素进行滤波处理,解决了经典空域滤波算法存在窗口内数据不满足滤波模型对同质区域要求的缺陷。同时,针对强散射点的特殊性,设计了相应的检测及处理方法。实验结果表明,该算法在获得与经典算法相当的相干斑抑制的同时,较经典算法具有更强的边缘和细节保持能力,同时获得更好的图像视觉效果。

When using the classical spatial filtering algorithm to process synthetic aperture radar （SAR） im ages, the identical sliding window is assumed to have homogeneous characteristics, which results in losing of image detail information. To solve the problem described above, a new sliding window structure is presented. The improved algorithm selects pixels that have similar statistical characteristics to the central pixel for struc- ture modeling. Meanwhile, a special detecting and processing strategy is designed to tackle strong scattering points. The airborne SAR image is processed to verify the effectiveness of the method. While maintaining the comparative ability in reducing speckle as the classical algorithm, the proposed algorithm achieves more edge and detail information as well as better image visual effects.