Cutset Sampling and Image Reconstruction

Cutset sampling is a new approach to acquiring 2D data where the
samples are taken densely along lines. This is different than typical
two-dimensional sampling, where a function is sampled on a rectangular
lattice. This talk focuses on a particular geometry of cutset
sampling called Manhattan-grid sampling, where the data is taken
densely along evenly spaced rows and columns. There are two main
motivations for this type of sampling: (a) there are physical
scenarios where it is natural, e.g. a ship sampling water temperature,
and (b) dense sampling along lines might capture sharp edge
transitions more completely than rectangular lattice sampling.