中文摘要：

针对多光谱图像中存在的多种噪声，提出一种利用空谱信息和分层字典学习的降噪方法．该方法依据相邻波段之间的结构相关性划分多光谱图像波段；并对得到的每个波段子集使用分层字典学习框架进行统计建模．通过引入高斯噪声项和稀疏噪声项，来有效地表达图像噪声特性；同时，应用吉布斯采样求解统计模型，以实现降噪的目的．在两幅真实多光谱图像数据上的仿真实验表明，该方法能够有效地抑制多光谱图像中的多种噪声，且能够准确地保留图像结构和细节信息．

英文摘要：

A novel denoising method is proposed for the multispectral irrlagery by combining the hierarchical dictionary learning and the spatial-spectral information. First, the band-subset segmentation is developed by exploiting the highly structural correlations between adjacent bands. Second, the hierarchical dictionary learning model with spatial information is applied to sequentially denoise each band-subset. The noise characteristics of the multispectral images is well depicted by decomposing the noise term into the Gaussian noise term and the sparse noise term, and Gibbs sampling is utilized to solve the model. The effectiveness of the proposed method is compared with that of the state-of-the-art approaches and validated on two multispectral images.