Abstract: In order to retain the edge information and remove the image noise as much as possible at the same time, a wavelet shrinkage algorithm was proposed, which took the inter-scale and intra-scale dependencies of wavelet coefficients into account. The proposed method used the correlation of wavelet coefficients and the average magnitudes of the surrounding wavelet coefficients within a local window to describe the inter-scale and intra-scale dependencies of wavelet coefficients, respectively. Thus, the image information and noise were identified. Meanwhile, a new threshold function was proposed to shrink wavelet coefficients. The experimental results show that the proposed denoising method can achieve high Peak Signal-to-Noise Ratio (PSNR).

Key words: image denoising, Gaussian noise, wavelet transform, wavelet threshold denoising, Peak Signal-to-Noise Ratio (PSNR)

摘要： 为了在保留图像边缘信息的同时，尽可能地去除图像噪声，提出一种基于小波系数尺度间和尺度内关系的去噪方法。该方法使用小波系数的相关系数和邻域小波系数的平均幅值来分别表示小波系数的尺度间和尺度内关系，并以此来辨别出图像的边缘信息和噪声;同时提出了一种阈值函数来处理图像的小波系数。实验表明该方法能取得较高的信噪比，并能保存图像的一些细节信息。

关键词: 图像去噪, 高斯噪声, 小波变换, 小波阈值去噪, 峰值信噪比