在变换域阚值去噪过程中,阈值的选取和阚值处理方法至关重要。提出一种基于conmurlet变换的图像去噪方法。采用分层阈值,为每一级contourlet系数选取一个阈值。阈值处理中给出一种基于邻域的阈值处理方法,不仅考虑单个系数幅值的大小,而且考虑它的邻域系数幅值的大小。同时为了抑制在去噪图像边缘附近的伪吉布斯效应,引入cycle spinning来抑制这种图像失真。实验结果表明,利用文中去噪方法进行去噪比其他方法得到更好的视觉效果和更高的PSNR值。
A method for image denoising based on the contourlet transform is proposed in this paper. Resolution - dependent thresholding is used to set a threshold for each stage of contourlet coefficients. A thresholding operator based on the neighborhood is introduced in the thresholding, considering not only the magnitude of a single coefficient but also the magnitudes of the coefficients in its neighborhood. Because the transform is shift - variant and leads to artifacts, cycle spinning is employed in this paper to restrain Gibbs- like phenomena near the image edges. The experimental results indicate that this method can get better visual effect and higher PSNR value than others.