Mercurial > hg > smallbox
annotate Problems/private/col2imstep.m @ 39:8f734534839a
(none)
author | idamnjanovic |
---|---|
date | Mon, 14 Mar 2011 15:35:01 +0000 |
parents | 51b76c31c93d |
children |
rev | line source |
---|---|
idamnjanovic@15 | 1 %COL2IMSTEP Rearrange matrix columns into blocks. |
idamnjanovic@15 | 2 % A = COL2IMSTEP(B,[MM NN],[N1 N2]) rearranges the columns of B into |
idamnjanovic@15 | 3 % sliding N1-by-N2 blocks producing the matrix A of size MM-by-NN. B is |
idamnjanovic@15 | 4 % usually the result of calling IM2COLSTEP(...) with a stepsize of 1, or |
idamnjanovic@15 | 5 % using Matlab's IM2COL(..,'sliding'). Overlapping blocks are summed in A. |
idamnjanovic@15 | 6 % |
idamnjanovic@15 | 7 % A = COL2IMSTEP(B,[MM NN],[N1 N2],[S1 S2]) arranges the blocks in A with |
idamnjanovic@15 | 8 % a step size of (S1,S2) between them. The first block is at A(1:N1,1:N2), |
idamnjanovic@15 | 9 % and the rest are at A((1:N1)+i*S1,(1:N2)+j*S2). Overlapping blocks are |
idamnjanovic@15 | 10 % summed in A. Note that B is usually the result of calling |
idamnjanovic@15 | 11 % IM2COLSTEP(...) with a stepsize of [S1 S2]. |
idamnjanovic@15 | 12 % |
idamnjanovic@15 | 13 % A = IM2COLSTEP(B,[MM NN KK],[N1 N2 N3],[S1 S2 S3]) generates a 3-D |
idamnjanovic@15 | 14 % output matrix A. The step size [S1 S2 S3] may be ommitted, and defaults |
idamnjanovic@15 | 15 % to [1 1 1]. |
idamnjanovic@15 | 16 % |
idamnjanovic@15 | 17 % See also IM2COLSTEP, IM2COL, COUNTCOVER. |
idamnjanovic@15 | 18 |
idamnjanovic@15 | 19 |
idamnjanovic@15 | 20 % Ron Rubinstein |
idamnjanovic@15 | 21 % Computer Science Department |
idamnjanovic@15 | 22 % Technion, Haifa 32000 Israel |
idamnjanovic@15 | 23 % ronrubin@cs |
idamnjanovic@15 | 24 % |
idamnjanovic@15 | 25 % August 2009 |