摘要
要非局部均值(non-local means,NLM)去噪算法已成为较有效去除图像噪声的算法之一。然而,当噪声水平较高时,NLM不能准确地计算图像块之间的相似度权重值,影响图像的去噪效果。针对上述问题,结合离散余弦变换(discrete cosinetransform,DCT)提出了基于DCT的非局部均值滤波算法。首先,利用DCT的低频系数重构图像,以达到滤除部分噪声的同时保护图像的主要内容。其次,利用重构图像较准确地计算图像块之间的相似度权重值,将NLM去噪算法用于噪声图像。实验结果表明,该算法能够得到较高的峰值信噪比(peak signal to noise ratio,PSNR)和更好的视觉效果。
Non-local means( NLM ) has been becoming one of the most useful tools for image denoising. How- ever, the calculation for its similarity weights has limited accuracy against noise when the noise level is too high. In order to handle above-mentioned problem, NLM denoising algorithm is introducd based on discrete cosine transform (DCT). First, making use of the low frequency of DCT to reconstruct image , the part of the denoise in image is filtered while preserving the main information of the image. Second, the NLM algorithm is used for the reconstruc- ted image to filter the additional denoise through calculation of similarity weights accurately. Compared with NLM algorithm, the experimental results demonstrate that our method gets a higher Peak Signal to Noise Ratio (PSNR) and better visual fidelity.
出处
《科学技术与工程》
北大核心
2013年第11期3123-3126,共4页
Science Technology and Engineering