基于多重分形的SAR图像降噪

SAR image denoising based on multifractal

调整奇异指数的值可以去除不规则的奇异点、保留有意义的奇异点,从而达到消噪的目的。理论上奇异指数的调整计算过于复杂,一般采用手工设定。手工设定的值带有很大的主观性、缺乏定量的描述,为此引入了合成孔径雷达(SAR)图像噪声抑制效果定量评判指标,通过条件迭代(ICM)的方法优选奇异指数值,优化了基于多重分形的图像降噪计算。实验结果表明相较于传统降噪方法,基于多重分形的方法不论是视觉效果还是定量评判指标都有较大的改善。

The noise can be reduced by adjusting the singular exponent to save regular singular points of the image. Theoretically singular exponent adjustment is so complex that commonly setting the singular exponent is handwork. However this exponent is subjective and lack of quantitative description. In this case the denois-ing effect evaluation indices have been introduced. The adjustment value is optimized through Iterated Condi-tional Mode, and image denoising calculation based on multifraetal is optimized too. Experiment results show that compared to conventional denoising algorithms, the visual effect and quantitative evaluation indices of this method are improved.