基于轮廓小波的图像信号去噪算法研究

Denoising Algorithm of Image Signal Based on Nonsubsampled Contourlet Transform

在非下采样Contourlet变换域中,针对憎水性图像相关特性,分析了图像有用信息与干扰噪声,提出了基于非下采样Contourlet变换复合绝缘子憎水性图像去噪算法,对变换后低频分量中含有光照不均匀成分采用B样条曲面进行近似,得到补偿后低频分量;对多分辨率多方向性带通分量中乘性噪声应用非线性扩散有选择滤波,最后对修正后系数重构。实验结果表明:与同态滤波相比,此算法不仅对憎水性图像光照不均匀部分最佳补偿,而且图像的细节、边缘信息得到有效的保留甚至加强,为后续憎水性图像分析与理解提供了良好的基础。

In the Nonsubsampled Contourlet Transform （NSCT） domain, and according to the relevant characteristics of the hydrophobic ima- ges, useful information and interference noise of the images was analyzed,with hydrophobic image denoising algorithm of composite insulator was proposed based on NSCT transform. B-splint surface approximation was used to uneven illumination composition in the low frequency. The muhiplicative noisein the band-pass multi-resolution multi-directional component was selective filtered by nonlinear diffusion. Then the modified coefficients were reconstructed finally. Compared with the homomorphic filtering, the experimental results show that：not only the best compensation of uneven illumination composition is obtained in the hydrophobic image, details and edge information of the image is effective to be retained or even strengthened. Solid foundation is provided for the follow-up of hydrophobic image analysis and understanding.