小波包与偏微分方程相结合的图像去噪方法

Image denoising algorithm based on wavelet packet and partial differential equations

图像在去噪时,梯度算子不能有效识别图像的灰度渐变区和图像淡边缘,而二阶微分量含有更丰富的信息,首先构建二阶微分算子,建立兼顾PM(Perona-Malik)模型和MCD(mean curvature diffusion)模型两者优点的权重函数,可自适应的去除噪声;再用小波包对噪声图像进行系数分解,克服权重函数易受噪声影响的弊端,建立基于小波包与偏微分方程的图像去噪算法。实验结果表明,该算法在兼顾保持区域内部较好光滑性的同时,很好地保持了边缘纹理信息,是一种理想的算法。

In image denoising,the image gray gradient zones and image light edges cannot be effectively identified by the gradient operators,while the quadratic differential contains more information. In this paper,the second order differential operators were established based on the advantages of both the Perona-Malik( PM) model and mean curvature diffusion( MCD) model,which can adaptively remove the noise. Then thewavelet packet is used for coefficients decomposition of noise images to overcome that the weight function is susceptible to noise. The image denoising algorithm based on wavelet packet and partial differential equations was then established. Experiment results show that the proposed algorithm can not only keep smoothness effect within the region,but also keep edge and texture information,which makes it an ideal algorithm.