一种基于轮廓波变换和各向异性扩散的图像去噪模型

AN IMAGE DENOISING MODEL BASED ON CONTOURLET TRANSFORM AND ANISOTROPIC DIFFUSION

提出一种采用轮廓波变换和各向异性扩散的图像去噪模型。利用轮廓波变换较好的稀疏性、多方向性等特点,通过对噪声图像经轮廓波变换后的不同尺度上的子带图像进行扩散,并采用P范数方法在轮廓波域计算子带图像的梯度阈值,实现建立在图像精细分析基础上的新的各向异性扩散模型。仿真结果表明,提出的扩散模型较好地抑制了传统各向异性扩散模型出现的边缘模糊效应,在对图像去噪的同时保留了更多的边缘、纹理等细节信息。

In this paper we propose an image denoising model which uses the Contourlet transform and anisotropic diffusion. By utilising the features of Contourlet transform in better sparsity and multi-directional property, and through diffusing the sub-band images of noise image on different scales with the Contourlet transform applied, as well as adopting P-norms method to calculate the gradient threshold of sub-band images in contourlet domain, we implement a new anisotropie diffusion model built on the basis of fine image analysis. Simulation results show that the edge-blurring effect in traditional anisotropic diffusion model can be well averted by this proposed model, and more detailed informa- tion such as edges and textures can be preserved while the image is denoising.