2021 Volume 56 Issue 3 Pages 116-123
The key of compressed sensing technology is to find out the feature elements (bases) contained in the original signal from few sampled measurements. In this paper, we introduce theories of compressed sensing techniques based on machine learning and their application to scanning transmission electron microscopy, electron energy loss spectroscopy, and electron holography, in which the bases are extracted by suitable statistical signal processing (sparse coding and tensor decomposition).