论文标题
用于压缩感测成像的预测性改进方法
Predictive refinement methodology for compressed sensing imaging
论文作者
论文摘要
弱 - $ \ ell^p $ norm可以用来定义稀疏的量子$ s $。当我们计算信号的离散余弦变换系数的$ s $时，$ s $的值与所述信号的信息内容有关。我们使用$ s $的此值来定义一个无参考索引$ \ Mathcal {e} $，称为稀疏索引，我们可以用来以高精度预测压缩感测成像的信号重建质量。这样，当压缩感测在抽样理论的上下文中时，我们可以使用$ \ Mathcal {e} $决定何时进一步分区采样空间并增加采样率，以优化我们使用压缩感测技术时的图像恢复。
The weak-$\ell^p$ norm can be used to define a measure $s$ of sparsity. When we compute $s$ for the discrete cosine transform coefficients of a signal, the value of $s$ is related to the information content of said signal. We use this value of $s$ to define a reference-free index $\mathcal{E}$, called the sparsity index, that we can use to predict with high accuracy the quality of signal reconstruction in the setting of compressed sensing imaging. That way, when compressed sensing is framed in the context of sampling theory, we can use $\mathcal{E}$ to decide when to further partition the sampling space and increase the sampling rate to optimize the recovery of an image when we use compressed sensing techniques.
