一种基于B样条双正交小波的图像阈值去噪方法

An Image Denoising Method Based on B-spline Biorthogonal Wavelet is Presented

针对图像中存在高斯噪声模糊图像,为提高图像的质量,利用二进提升方案与小波多分辨率分析的优势,提出一种基于B样条双正交小波的图像阈值去噪方法。首先利用二进提升方案构造一个新的双正交小波滤波器,对含噪图像进行小波变换,选择新构造的双正交小波和适当的分解层次,得到低频子图像和水平、垂直、对角线三个高频子图像。对每一次分解得到的高频子图像在不同尺度上使用不同的阈值模型进行去噪,然后对低频子图像与去噪后的所有高频子图像进行小波重构,得到最终去噪后的图像。

In order to improve the quality of the image,a threshold denoising method based on B-spline bispectral wavelet is proposed by using the advantages of binary lifting scheme and wavelet multi-resolution analysis.Firstly,a new biorthogonal wavelet filter is constructed by using dyadic lifting scheme,and the noisy image is transformed by wavelet transform.The low-frequency sub image and three high-frequency sub-images of horizontal,vertical and diagonal are obtained by selecting the newly constructed bisex wavelet and appropriate decomposition level.The high-frequency sub images obtained by each decomposition are denoised with different threshold models at different scales Then the low-frequency sub image and all high-frequency sub-images obtained after denoising are reconstructed by wavelet to get the final denoised image.