一种快速交叠组合稀疏全变分图像去噪方法

A Fast Image Denoising Method Based on Overlapping Group Sparsity Total Variation

交叠组合稀疏全变分(Overlapping Group Sparsity Total Variation,OGSTV)是一种能够比较有效地克服图像去噪过程中产生“阶梯伪影”问题的模型,但其求解方法在图像去噪性能和处理时间上仍存在一定的提升空间.本文在OGSTV模型基础上,提出一种利用快速傅里叶变换(FastFourierTransform,FFT)方法对SplitBregman求解算法进行优化的快速OGSTV图像去噪方法.实验结果表明:在采用SplitBregman优化算法的OGSTV模型中引入FFT后,不仅绝大部分提高了OSGSTV的图像去噪性能,而且明显减少了OGSTV对图像进行去噪所需的时间.

The overlapping group sparsity total variation (OGSTV) is a more effective model to overcome the “staircase artifacts” emerged in the procedure of image denoising. But the performance and cost-time can be more enhanced when the traditional solution method is used. By introducing fast Fourier transform for split Bregman algorithm applied in OGSTV, a fast algorithm is proposed to solve the OGSTV model. The experimental results demonstrate that, after introducing FFT in OGSTV, almost all the quality of images denoising is better than ever, but also the denoising time is reduced obviously.