PGAUSS |
Vizualizes set of bivariate Gaussians.
Synopsis:
pgauss(model)
pgauss(model,options)
h = pgauss(...)
Description:
pgauss(model) visualizes a set of bivariate Gaussians as
isolines (ellipse) with equal probability density functions.
The Gaussians are given by mean vectors model.Mean [2xncomp]
and covariance matrices model.Cov [2x2xncomp]. If labels
model.y [1xncomp] are given then the Gaussians are distinguished
by colors correspoding to labels.
pgauss(model,options) structure options controls visualization;
If options.fill equals 1 then Ellipses are filled otherwise only
contours are plotted. The isolines to be drawn are given by
values of probability distribution function in field
options.p [1xncomp]. If length(option.p)==1 then isolines for
all Gaussians are drawn for the same value.
h = pgauss(...) returns handles of used graphics objects.
Input:
model [struct] Parameters of Gaussian distributions:
.Mean [2 x ncomp] Mean vectors of ncomp Gaussians.
.Cov [2 x 2 x ncomp] Covariance matrices.
.y [1 x ncomp] (optional) Labels of Gaussians used to distingush
them by colors. If y is not given then y = 1:ncomp is used.
options.p [1 x ncomp] Value of p.d.f on the draw isolines.
If not given then p is computed to make non-overlapping isolines.
options.fill [int] If 1 then ellipses are filled (default 0).
Output:
h [1 x nobjects] Handles of used graphics objects.
Example:
data = load('riply_trn');
model = mlcgmm( data );
figure; hold on;
ppatterns(data);
pgauss( model );
About: Statistical Pattern Recognition Toolbox
(C) 1999-2003, Written by Vojtech Franc and Vaclav Hlavac
Czech Technical University Prague
Faculty of Electrical Engineering
Center for Machine Perception
Modifications:
23-aug-2004, VF, uses model.y to color plots in 1D case
30-apr-2004, VF