基于小波域的维纳滤波算法在锥束牙科CT中的应用

摘要

A wavelet domain Wiener filtering method is proposed for denoising of cone-beam dental CT images.The db4 wavelet is used to decompose the CT image,and the Wiener filtering is implemented in each scale of wavelet coefficients,while the parameters of local mean and noise variance estimation are adjusted based on region statistical characteristics in each scale of wavelet coefficients.New wavelet coefficients are used for the reconstruction of the denoised image,and the denoised CT image is obtained.The results of computer simulation and the cone beam dental CT actual data prove that this method raises the reconstruction image's signal-to-noise ratio,suppresses image noise effectively,and then improves the image visual effect obviously.%提出一种小波域的维纳滤波方法对锥束牙科CT断层图像进行降噪.该算法以db4小波作为分解小波对CT图像进行分解,在分解后的每个子带再进行维纳滤波,并根据图像的区域统计特性对每个子带的局部均值和噪声方差估计参数进行了调整.利用降噪后的小波系数重构图像,得到降噪后的CT断层图像.通过计算机仿真及锥束牙科CT的真实数据测试表明,本文采用的方法有效抑制了图像噪声,提高了图像的信噪比,明显改善了图像的视觉效果.