论文部分内容阅读
针对高光谱遥感图像(Hyperspectral Image,HSI)去噪问题,提出了基于非局部低秩字典学习的图像去噪算法。该算法利用高光谱遥感图像各波段之间的强相关性,结合图像非局部自相似性和局部稀疏性提高去噪性能。首先,结合各波段图像的强相关性、非局部自相似性和局部稀疏性建立非局部低秩字典学习模型,然后,利用迭代法求解该模型得到冗余字典和稀疏表示系数,最后,利用冗余字典和稀疏表示系数复原图像。相比较现有先进的算法,由于充分利用了高光谱图像各波段的强相关性这一内在特征,使得该算法能够很好地保持高光谱遥感图像的细节信息,达到了预期效果。
To solve the problem of HSI (Hyperspectral Image) denoising, an image denoising algorithm based on non-local low-rank dictionary learning is proposed. The algorithm utilizes the strong correlation between the bands of hyperspectral remote sensing images, and combines the non-local self-similarity and local sparsity to improve the denoising performance. Firstly, a non-local low-rank dictionary learning model is constructed based on the strong correlation, non-local self-similarity and local sparsity of each band image. Then the iterative method is used to solve the model to obtain redundant dictionaries and sparse representation coefficients. Finally, I dictionaries and sparse representation coefficients to reconstruct the image. Compared with the existing advanced algorithms, the intrinsic feature of the strong correlation of the bands of the hyperspectral image is fully utilized, so that the algorithm can well maintain the detail information of the hyperspectral remote sensing image and achieve the expected result.