【摘要】 针对目前大多图像去噪算法的性能依赖输入噪声水平参数的问题,为进一步提高去噪效果,提出一种改进的基于非局部相似和低秩模型的图像盲去噪方法。预先估计图像的全局噪声方差,在图像非局部相似和低秩模型的框架下,自适应地估计各图像块的局部噪声方差,确定各图像块奇异值阈值(SVT)的局部阈值参数,运用迭代规则完成去噪。为验证该方法的有效性,与3种目前较成熟的去噪算法进行仿真对比。仿真结果表明,对于噪声方差未知的图像,该方法的去噪效果在视觉、峰值信噪比(PSNR)和结构相似度(SSIM)的数据上更具优势,具有更好的自适应能力,更适合应用于实际图像去噪问题。更多还原

【Abstract】 Considering the requirement for the noise level parameter of the majority of existing image denoising algorithms,to improve the denoising quality,a blind image denoising algorithm based on nonlocal similarity and low rank model was proposed.The image’s global noise variance was estimated in advance.Within the framework of the image’s nonlocal similarity and low rank model,image patches’ local noise variance was adaptively estimated to get these image patches’ local threshold parameter of the singular value thresholding(SVT),and iteration regularization was utilized to complete image denoising.To verify the validity of this algorithm,it was campared with three state-of-theart denoising algorithms.Experimental results indicate that the proposed method is better than the compared methods on both visual effects and the performance of PSNR and SSIM for the image with unknown noise variance.The proposed method has better adaptive ability and is more suitable for the practical image denoising.更多还原