为了提高固体二维双量子-单量子(DQ-SQ)谱的采集速度，根据DQ-SQ谱的自稀疏性，该文使用了一种基于压缩感知技术的重建算法。其优化的能量函数是有限差分约束的l1范数，并使用不同的权重对水平和竖直方向的有限差分项进行约束。该文分别对伪随机采样、全随机采样和e指数采样等采样模式进行了比较，发现伪随机采样表现出最佳的重建结果。进一步研究发现伪随机的极限形式，即t1截尾(t1-cutoff)采样模式效果最佳。%To increase the speed of acquisition of two-dimensional solid-state DQ-SQ spectrum, a compressed sensing algorithm which makes use of the self-sparsity of the spectrum to construct under-sampled data. The energy function used in optimization is l1 norm together with the finite difference term. In the finite different term, we used different weights for the horizontal and vertical finite differences. Different sampling schemes were compared and pseudo-random sampling combined with compressed sensing reconstruction was found to yield the best results. Furthermore, we found that the extreme case of pseudo-random sampling, that is, t1-cutoff sampling may be the best choice.
展开▼
