基于非采样Contourlet与全变差模型的图像去噪

Half-Soft Threshold Image De-noising Algorithm Based on Nonsubsampled Contourlet Transform and Total Variation

分析非采样Contourlet与全变差滤波器在图像去噪中的特点,提出一种基于图像全变差模型的非线性扩散与非采样Contourlet相结合的自适应混合图像去噪算法.该算法在中、低频部分采用全变差扩散,在高频部分采用非采样Contourlet变换,并在此基础上用半软阈值法取代软阈值法.实验结果表明,该算法能够在降低复杂度的同时得到好的滤波效果.

The features of nonsubsampled Contourlet transform（NSCT）and total variation （TV） in image de-noising were studied. An adapted hybrid image de-noising algorithm is proposed based on TV diffusion and NSCT. The TV diffusion was applied to the medium and low frequency parts of image decomposed by NSCT. A half-soft threshold other than the soft threshold was used to shrink the high frequency parts. Experiment results show that the de-noising performance was improved by the proposed algorithm.