<div>Ok as far as I can see it's</div>
<div>Category ID (class)</div>
<div>Mean vector (each value is the average for the category in each band)</div>
<div> </div>
<div>Var(b1;b1)</div>
<div>Cov(b1,b2) Var(b2;b2)</div>
<div>Cov(b1,b3) Cov(b2,b3) Var(v3,v3)</div>
<div><br>it's this?</div>
<div>By the way, is in any place the inversion of the covariance matrix?</div>
<div> </div>
<div>Thanks</div>
<div>Pedro<br><br></div>
<div class="gmail_quote">On Wed, Jul 28, 2010 at 8:30 PM, Nikos Alexandris <span dir="ltr"><<a href="mailto:nikos.alexandris@felis.uni-freiburg.de">nikos.alexandris@felis.uni-freiburg.de</a>></span> wrote:<br>
<blockquote style="BORDER-LEFT: #ccc 1px solid; MARGIN: 0px 0px 0px 0.8ex; PADDING-LEFT: 1ex" class="gmail_quote">
<div class="im">On Wednesday 28 of July 2010 21:22:08 Nikos Alexandris wrote:<br>> Perhaps just reading the created signature (text) files and identifying<br>> the mean's and variance/covariance matrices is enough.<br>
<br></div>Didn't have a look yet but the "Mean's ... matrices" is I guess wrong. The<br>signature file probably contains mean values (one mean value for each training<br>class) and the var./covar. matrices (one matrix for each training class).<br>
<font color="#888888"><br>Nikos<br></font></blockquote></div><br>