论文部分内容阅读
针对激光遥感图像的超分辨重建问题,本文提出了基于联合稀疏字典学习的单幅图像超分辨重建算法,并将其用于激光遥感图像的超分辨重建。该算法首先利用已有的高分辨图像,通过预处理得到高低分辨样本集。然后提出了联合稀疏字典学习技术,并利用该技术对高低分辨样本集训练得到稀疏的高低分辨字典。最后利用高低分辨字典重建高分辨图像,并分析了算法的计算复杂度。相比较现有的激光遥感超分辨重建算法,由于采用了联合稀疏字典学习技术,本文提出的算法训练字典时需要较少的样本数和计算量,同时由于稀疏字典学习能够提高数据表示的精度,重建效果得到了进一步的提高。实验结果验证了算法的有效性。
For super-resolution reconstruction of laser remote sensing images, this paper proposes a single image super-resolution reconstruction algorithm based on joint sparse dictionary learning, which is used to super-resolution reconstruction of laser remote sensing images. The algorithm first uses the existing high resolution image to get the high and low resolution sample set through preprocessing. Then, a joint sparse dictionary learning technique is proposed, and a sparse high-resolution dictionary is trained by using the technique. Finally, the high-resolution dictionary is reconstructed using high and low resolution dictionaries, and the computational complexity of the algorithm is analyzed. Compared with the existing laser remote sensing super-resolution reconstruction algorithm, the algorithm proposed in this paper needs fewer samples and computational resources to train the dictionary because of the sparse dictionary learning technology. Meanwhile, because sparse dictionary learning can improve the accuracy of data representation, The reconstruction effect has been further improved. Experimental results verify the effectiveness of the algorithm.