[Insight-users] GeodesicActiveContour problem
Jayant Chauhan
gripened at hotmail.com
Thu, 22 Apr 2004 14:39:53 +0530
<html><div style='background-color:'><DIV class=RTE>Dear Luis,</DIV>
<DIV class=RTE><BR>As I had written to you earlier, I am still stuck with my initial level<BR>set breakinginto more contours, which then expand spread throughout the<BR>image. I have already sent you the code.</DIV>
<DIV class=RTE> Could it be a problem because my initial level set is a rectangle ?? Have<BR>the cases for SHOCKS been implemented ?? It really beats me as to why the<BR>contour is expanding, because outside the rectangle each pixel is set to<BR>ZERO (is infact zero). ALso I have done :</DIV>
<DIV class=RTE> typedef itk::GeodesicActiveContourLevelSetImageFilter<InternalImageType,<BR> InternalImageType> GACFilterType;<BR> GACFilterType::Pointer gAC = GACFilterType::New();</DIV>
<DIV class=RTE> gAC->SetReverseExpansionDirection(true); // Is this correct ??</DIV>
<DIV class=RTE>I dont know why this is happening.<BR>After alot of iterations, say 10000, I get one contour wroughly around the<BR>object and another contour which is almost at the boundary of the image.<BR>And it is the second contour which is resulting in so many iterations.</DIV>
<DIV class=RTE> </DIV>
<DIV class=RTE>2) ALso, is there any other way of getting/creating an initial estimate about the moving object other than taking the difference between two consecutive frames ??</DIV>
<DIV class=RTE> </DIV>
<DIV class=RTE>with regards</DIV>
<DIV class=RTE>Jayant Chauhan</DIV></div><br clear=all><hr>BharatMatrimony.com. ? <a href="http://g.msn.com/8HMAENIN/2740??PS=">The No 1 Indian matrimony.</a> </html>