A Parametric Texture Model Based on Joint Statistics of Complex Wavelet Coefficients. Portilla, J. and Simoncelli, E., P. International Journal of Computer Vision, 40(1):49-70, 2000.
bibtex   
@article{
 title = {A Parametric Texture Model Based on Joint Statistics of Complex Wavelet Coefficients},
 type = {article},
 year = {2000},
 pages = {49-70},
 volume = {40},
 id = {8c4932d7-17b7-3b3d-adca-b201f52a8936},
 created = {2015-02-26T19:48:01.000Z},
 file_attached = {false},
 profile_id = {624218fb-8957-3e79-afb5-27cca967200e},
 group_id = {ac791549-4a16-3252-8de1-0fdef1a59e25},
 last_modified = {2015-03-10T17:06:55.000Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {true},
 hidden = {false},
 citation_key = {Portilla:2000gl},
 source_type = {article},
 notes = {Parameters and correspondings variables in params struct returned by textureAnalysis.m<br/><br/>marginal stats-----<br/>skew &amp;amp; kurtosis: statsLPim<br/>high-pass band: vHPR0<br/><br/>raw coefficient correlation-----<br/>central samples ( (MxM+1)/2, rotationally symmetrically stored in a MxM matrix) of the auto correlation of low pass images: autoCorreReal<br/><br/>coefficient magnitude stats-----<br/>central samples of auto-correlation of magnitude of each subband ( (MxM+1)/2, rotationally symmetrically stored in a MxM matrix) : autoCorrMag<br/>cross-correlation of each subband at the same scale: C0<br/>cross correlation of magnitudes with coarser scale: Cx0<br/><br/>cross-scale phase statistics----<br/>real with both the real and imaginary of phase-doubled coefficients at next coarser scale: Crx0<br/><br/>some parameters are not mentioned in the paper. Refer to the commented code.<br/><br/>statg0 = params.pixelStats; % all by cmask(1)<br/>mean0 = statg0(1); <br/>var0 = statg0(2);<br/>skew0 = statg0(3); <br/>kurt0 = statg0(4);<br/>mn0 = statg0(5); <br/>mx0 = statg0(6);<br/>statsLPim = params.pixelLPStats;<br/>skew0p = statsLPim(:,1); % by cmask(1)<br/>kurt0p = statsLPim(:,2); % by cmask(1)<br/>vHPR0 = params.varianceHPR; % by cmask(2)|cmask(3)|cmask(4)<br/>acr0 = params.autoCorrReal; % by cmask(2)<br/>ace0 = params.autoCorrMag; % by cmask(3)<br/>magMeans0 = params.magMeans; % not changed?<br/>C0 = params.cousinMagCorr; % by mask(3)<br/>Cx0 = params.parentMagCorr; % by mask(3)<br/>Crx0 = params.parentRealCorr; % by cmask(4)<br/><br/><br/>% cmask (optional): binary column vector (4x1) indicating which sets of<br/>% constraints we want to apply in the synthesis. The four sets are:<br/>% 1) Marginal statistics (mean, var, skew, kurt, range)<br/>% 2) Correlation of subbands (space, orientation, scale)<br/>% 3) Correlation of magnitude responses (sp, or, sc)<br/>% 4) Relative local phase},
 folder_uuids = {886c64ff-fda7-4525-b6b0-c7319eef9f2d,db2540d2-49f3-4068-8c93-f82b72c33309},
 bibtype = {article},
 author = {Portilla, Javier and Simoncelli, Eero P},
 journal = {International Journal of Computer Vision},
 number = {1}
}
Downloads: 0