图像线状和网格状噪声的去除方法

Method for Denoising Linear and Gridded Image

在叠加了线状和网格状噪声图像的频谱中,噪声分布具有准周期性且对应于坐标轴上的亮点,由此提出了基于陷波滤波器的噪声去除算法。先对带有线状和网格状噪声的图像进行傅里叶变换,检测亮点所在位置,再采用陷波滤波器滤波,然后对滤波结果进行傅里叶逆变换,得到去噪声图像。为了取得更好的去噪效果,在空域中使用了开关中值滤波以及退化模型进行图像复原,仿真结果表明复原效果比采用陷波滤波器进行滤波还原更好。

The noise distribution in linear and gridded noise image is quasi periodic and the noise corresponds to the bright spots in its Fourier spectrum.According to this principle,the paper proposed denoising algorithm by using notch filtering.First,the linear and gridded noise image was carried out Fourier transform.Second,the location of the bright spots was detected and filtered by using notch filter.Finally,the restored image was obtained after the inverse Fourier transform of the filtering results was performed.On this basis,for achieving better denoising effect,the switch median filtering and devolution model were adopted to remove the noise in the spatial domain.The simulation results showed that in the spatial domain it would achieve better denoising effect to adopt switch median filtering and devolution model to restore image than to adopt the notch filtering.