基于字典学习和原子聚类的图像去噪算法

针对图像去噪过程中会导致细节和纹理结构信息丢失的不足,本文提出了基于字典学习和原子聚类的图像去噪算法。该算法首先利用含噪图像通过字典学习算法得到自适应的冗余字典,然后提取字典中每个原子的HOG特征和灰度统计特征构成特征集,并利用原子的特征集将冗余字典中的原子分成两类(不含噪原子和噪声原子),最后利用不含噪原子恢复图像,达到去噪的目的。实验结果表明,本文提出的算法无需知道噪声的先验信息,峰值信噪比好于现有的流行算法,且能较好地保持图像细节和纹理结构信息,提高了视觉效果。

Image Denoising Based On Dictionary Learning And Atom Clustering

For the shortcoming of losing texture structures with image denoising process, the image denoising algorithm based on dictionary learning and atom clustering. Firstly, adaptive redundant dictionary is obtained by noised image over dictionary learning. Then the set of features is structured by HOG features and histogram of gray features, and atoms of redundant dictionary is clustered to two classes by the set of features. Finally, image is restored by denoised atoms and noise is removed. Our experimental results demonstrate that the prior knowledge of the noise is unnecessary and the peak signal to noise ratio Value (PSNR) of the proposed algorithm is better than state-of-the-art algorithms, while the proposed algorithm can well preserved the texture structures in the denoised image, making them look more natural.