Figure 1
From: An augmented Lagrangian multi-scale dictionary learning algorithm
A two-dimensional toy example testing the atom identify ability of K-SVD algorithm for two different sample types with the size of 1,000. Each sample is a multiple of one of eight basis vectors plus additive noise. (a) The dictionary's atoms have a uniform angular distribution in the circle; (b) Seven atoms have a uniform angular distribution in half a circle except the last one. A display of one run is shown above, where the green vectors show the true atoms and the red vectors show the learned atoms.