摘要： 为提高方向小波变换方向选择的自适应性，提出一种混合傅里叶变换和方向小波变换的图像去噪算法。结合图像的边缘信息，利用傅里叶变换估算图像的局部方向，构造方向小波变换的生成矩阵。考虑方向小波变换陪集分解的特点，根据序列长度选择相应阈值。实验结果表明，当噪声方差大于25时，该方法的去噪性能较好。

关键词: 图像去噪, 方向小波变换, 生成矩阵, 傅里叶变换, 局部方向信息

Abstract: To solve the non-adaptive choice of direction of the directionlet transform, a new hybrid Fourier-directionlet transform denoising method based on local direction information is proposed. It uses Fourier transform to estimate the local direction according to the edge information of the image and deals with the local direction as the generator matrix of directionlet transform. According to the characteristics of the directionlet transform, the pixel sequences are divided into three sets and dealt with different threshold method respectively. Experimental results show that the proposed method can remove the noise more effectively when the variance of noise is greater than 25.

Key words: image denoising, directionlet transform, generator matrix, Fourier transform, local direction information

中图分类号: 

• TP391