基于多重分形分析的SAR图像消噪增强研究

On Denoising SAR Image by Processing Based on Multifractal Analysis

根据 SAR图像中原始信号与噪声的多重分形谱的不同 ,提出了一种基于奇异性分析的雷达图像消噪新方法。在图像经过多重分形分析后 ,通过对多重分形谱相关的 Holder指数进行特征化处理来实现图像消噪。经过实验确定当变化后的 Holder指数略大于 1 .7时 ,最终恢复的图像消噪效果最佳。这种方法可以在尽量保留信号奇异性 (如边缘 ,纹理等 )相关信息情况下 ,使得图像平滑 。

To our best knowledge, we are the first in China to denoise SAR(Synthetic Aperture Radar) image by processing based on multifractal analysis. The multifractal spectrum of the original signal is different from that of noise. We use this difference to denoise SAR image . Section 3 discusses in detail how to denoise SAR image by processing based on multifractal analysis. Essentially our method focuses on adjusting the Hlder exponent α of multifractal spectrum. After much simulation research, we find that α should be adjusted to 1.72~1.73. The higher the value of α is above 1.73, the less distinct the edges of SAR image become. In our opinion, denoising is optimum at α =1.72~1.73. In other words, when α =1.72~1.73, we obtain a smooth and denoised SAR image.