
Full text loading...
m = W ⋅ G m = (G H H ⋅ (G ⋅ W ⋅ G 2 ⋅ G + λ W −1 −1 ) H 2 + λ I) ⋅ d (E 2) ⋅ G ⋅ d (E 3) The weighting matrix W is the key point. If it is the identity-matrix we end up with the standard least-squares decomposition. To obtain a high resolution decomposition some weights are put on the diagonal to enforce the focusing of the decomposition on its most significant spectral components. For example: H −1 ( ) ( ) ( ) ⎟ ⎠ ⎜