how much is your physical mem available
(128/256/512/1/2)?? my personal experience with 512
megs of mem...make an haarclassifiercascade.txt until
stage 27 is smoothly executed with 256 megs of mem
allocated...the program slow down if you set a memory
above your physical memory (i have try with -mem 768)
and for 1 stage require 10 minutes to complete

you sure had follow the right procedures??

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

original:in the last lines of function BlobAnalysis(), the following
code
will cause "vector iterators incompatible" error as
the erase operation invalidates the itBlobs,though you may use
"#define _HAS_ITERATOR_DEBUGGING 0" to prevent assertion,the itBlobs
may point
to incorrect address in momery, as the change of vector's size may
move the whole
vector to a new address,so the modified version may correct
this,there's another
place in the source causes the same trouble,use the same way may
resolve it:)
// eliminem els blobs subsumats
	 blob_vector::iterator itBlobs = RegionData.begin() +
HighRegionNum + 1;
	 while( itBlobs != RegionData.end() )
	 {
		 delete *itBlobs;
		 RegionData.erase(itBlobs);

	 }

modified:
// eliminem els blobs subsumats
	 blob_vector::iterator itBlobs = RegionData.begin() +
HighRegionNum + 1;
	 while( itBlobs != RegionData.end() )
	 {
		 delete *itBlobs;
		 itBlobs++;

	 }
	 itBlobs = RegionData.begin() + HighRegionNum + 1;
	 if( itBlobs != RegionData.end() )
	 {

		 RegionData.erase( itBlobs,RegionData.end() );
	 }


another place:
original in filter() in BlobResult.cpp
	 if( &dst == this )
	 {
		 // esborrem els primers blobs ( que són els
originals )
		 // ja que els tindrem replicats al final si
passen el filtre
		 blob_vector::iterator itBlobs = m_blobs.begin();
		 for( int i = 0; i < numBlobs; i++ )
		 {
			 delete *itBlobs;
			 m_blobs.erase(itBlobs);
		 }



	 }

modified:
	 if( &dst == this )
	 {
		 // esborrem els primers blobs ( que són els
originals )
		 // ja que els tindrem replicats al final si
passen el filtre
		 blob_vector::iterator itBlobs = m_blobs.begin();
		 for( int i = 0; i < numBlobs; i++ )
		 {
			 delete *itBlobs;
			 itBlobs++;
		 }
		 itBlobs = m_blobs.begin();
		 if (numBlobs>0)
		 {
			 m_blobs.erase( itBlobs,m_blobs.end() );
		 }


	 }

I'll try to be more precise:

In our tests, SIFT extraction was too slow for real time, so as a
result the tracker gets lost as we can't process enough frames per
second.
On the other hand, Harris corner extractor is fast enough (we've
reached 15 fps in good conditions, with acceptable navigation, but
not stable enough), but most
of the Harris corners are features that are not stable over time, as
lots of them are light reflexions and stuff like this, so when the
robot moves he doesn't find the same features again and it's hard to
get stability.
Some people suggests us to use artags, but the goal was to navigate
in an unknown environment...
thanks

darío

--- In OpenCV@yahoogroups.com, "Peaceful Monkey"
<anthony.remazeilles@...> wrote:
>
> Could you be precise when you say that the harris corner are not
stable
> for navigation ? Are you talking about the feature tracker or the
> feature extractor ?   Cordelia Schmid
> <http://lear.inrialpes.fr/people/schmid/publications.html>  did a
really
> good job showing the property of different feature extractor. KLT
> tracker has been widely use within so many applications....
>
> Extraction and tracking (and even matching) are not the same thing.
It
> is not clear to understand which part you would like to improve in
your
> application
>
> Thanks
>
> Anthony
>
> --- In OpenCV@yahoogroups.com, "darioroto" <darioroto@> wrote:
> >
> > hi,
> >
> > we're a group that is working on a humanoid robot, and we're
trying to
> > implement a visual SLAM program to navigate in an unknown
environment.
> >
> > Has anybody tried to implement SIFT landmark detection in openCV?
> > We've tried with Harris corners and also with eigenvalues (both
from
> > cvGoodFeaturesToTrack) but they're not stable enough to navigate.
Has
> > anybody tried SIFT landmarks for navigation? We haven't
implemented it
> > yet, but the few examples that we've found on the net where
> reeeeeeally
> > slow, so we can't try them, as we need something to navigate in
real
> > time...
> >
>
>
>
> [Non-text portions of this message have been removed]
>

Hi,

Please, somebody worked with Delaunay triangulation in OpenCV?

I need extract triangles from a set point in a image, but in OpenCV
return only the edges.

Regards,

Fábio S. Vidal

Dear All,

I want to grab from a frame grabber using directshow. I found that
cvcam  uses directshow to do this, but I could not find
"iproxytrans.h" anywhere and I could not compile the cvcam. do you
know how to compile cvcam source and/or how to use directshow to grab
(an easy way please!)?

Regards,
Mohsen

_____

From: Fadi Chakik [mailto:chakik@...]
Sent: Tuesday, January 02, 2007 9:14 AM
To: 'opencv@yahoogroups.com'
Subject: cvSVM: Support Vector Machines.



Hello,

Is it possible to send a working example code (training and testing) on how
to use one of the builtin  opencv machine learning ( cvSVM (Support Vector
Machines) or Random Trees or Decision Trees are)?

Thank you in advance for your help.

-Fadi

Email: chakik@...





[Non-text portions of this message have been removed]

I have tried several codecs to write .avi files and play them after, may be
this is because I'm using FC6, but the only codec that really worked was
"xvid". Most of them write the file but ffmpeg, or mplayer, are unable to
playback the videos.
Does anybody knows why this kind of things happen?

Victor Benso

On 12/23/06, Dong Si Tue Cuong <dongsi_tuecuong@...> wrote:
>
>   Hi,
> Stephen: Thanks for your attention. By the time I tried to run, it is "
> out.avi". I just desperately changed the name to see any hope.
>
> Anwyay, I found the cause. I post here for those who have the same error:
> The reason is that the VideoWriter is null, which comes from unsupported
> codec in OpenCV. When tracing back, I found that I install OpenCV before
> installing ffmpeg and ffmpeg support is unchecked when running OpenCV's
> "./configure".
> To fix, just reinstall OpenCV. Make sure ffmpeg support is checked "yes".
>
> However, there is still a problem. The output video file cannot be
> replayed. The same program runs fine on Windows, i.e. the output can be
> replayed.
> Any help would be much appreciated.
> Thanks for reading.
> Regards,
>
>
> Stephen DiVerdi <stephen.diverdi@... <stephen.diverdi%40gmail.com>>
> wrote:
>
> this may be a silly suggestion, but in your comment you say you're looking
>
> for "out.avi" but in your code you're creating a video called "sample.avi"
>
> - are you sure the output isn't there?
>
> -stephen diverdi
> -sdiverdi@... <-sdiverdi%40cs.ucsb.edu>
>
> On Thu, 21 Dec 2006, Dong Si Tue Cuong wrote:
>
> > Hi,
> > I was trying to write images captured from a FireWire camera (Bumblebee)
> into a video file. I am using CvVideoWriter. The program runs smoothly and
> no error is reported. But the output file "out.avi" is no where.
> > Before that, I tried to save captured images into ".ppm" files and it
> worked correctly. Before putting each frame into the output video file, I
> display the current image and they are all fine. But the output "out.avi"
> is missing.
> > I tried using different codecs (fourcc input) but it doesn't work.
> > Please help.
> > The part of video writing is as follow. Sorry for the worng indentation
> when copying from my editor into this. The idea of the code is: after
> initialization, loop to do the capture, Bayer conversion, put pixel info
> into the IplImage structure and then write frame into video file, for
> FRAME_NUM times.
> >
> > /*------------------- start of video writing
> --------------------------------*/
> > // OpenCV variables
> > IplImage* img = cvCreateImage( cvSize( nCols, nRows ), IPL_DEPTH_8U, 3
> );
> > CvVideoWriter* writer = cvCreateVideoWriter( "sample.avi", CV_FOURCC(
> 'D', 'I', 'V', 'X' ), 12.0, cvSize( nCols, nRows ) );
> > int height = img->height; // nRows
> > int width = img->width; // nCols
> > int step = img->widthStep;
> > int channels = img->nChannels;
> > uchar_t* data = (uchar_t *) img->imageData;
> > int i1, i2;
> >
> > //create a window
> > cvNamedWindow( "displayWindow", CV_WINDOW_AUTOSIZE );
> >
> > for ( count = 0; count < FRAME_NUM; count++ ) {
> >
> > // capture one frame
> > dc1394_video_set_one_shot( camera, DC1394_ON );
> > if ( dc1394_capture_dma( &camera, 1, DC1394_VIDEO1394_WAIT )!=
> DC1394_SUCCESS)
> > {
> > fprintf( stderr, "Unable to capture a frame\n" );
> > cleanup_and_exit( camera );
> > }
> > pBayerImage = dc1394_capture_get_dma_buffer( camera );
> > dc1394_capture_dma_done_with_buffer( camera );
> >
> > // do the Bayer conversion
> > dc1394_bayer_decoding_8bit( pBayerImage,
> > pRGBImage,
> > nCols,
> > nRows,
> > bayerPattern,
> > DC1394_BAYER_METHOD_NEAREST );
> >
> > // copy to IplImage data structure
> > for ( i1 = 0; i1 < height; i1++ ) {
> > for ( i2 = 0; i2 < width; i2++ ) {
> > data[i1*step+i2*channels+2] = pRGBImage[i1*step+i2*channels];
> > data[i1*step+i2*channels+1] = pRGBImage[i1*step+i2*channels+1];
> > data[i1*step+i2*channels] = pRGBImage[i1*step+i2*channels+2];
> > }
> > }
> >
> > // display image
> > cvShowImage( "displayWindow", img );
> > printf( "capture image %i\n", count );
> > cvWriteFrame( writer, img );
> > cvWaitKey( 1000 );
> > }
> >
> >
> > __________________________________________________
> > Do You Yahoo!?
> > Tired of spam? Yahoo! Mail has the best spam protection around
> > http://mail.yahoo.com
> >
> > [Non-text portions of this message have been removed]
> >
> >
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam? Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
>
> [Non-text portions of this message have been removed]
>
>
>


[Non-text portions of this message have been removed]

Sorry,the second should be the following
> modified:
>  if( &dst == this )
>  {
> 	 // esborrem els primers blobs ( que són els
> originals )
> 	 // ja que els tindrem replicats al final si
> passen el filtre
> 	 blob_vector::iterator itBlobs = m_blobs.begin();
> 	 for( int i = 0; i < numBlobs; i++ )
> 	 {
> 		 delete *itBlobs;
> 		 itBlobs++;
> 	 }
> 	 itBlobs = m_blobs.begin();
> 	 if (numBlobs>0)
> 	 {
> 		 m_blobs.erase( itBlobs,itBlobs+numBlobs);
> 	 }
>
>
>  }

but as the release of OpenCV 1.0, some old code will  run-time errer
with the new libs and dlls,considering rewriting the code without
cvblobslib,because any code without adequate documentation and
examples will bug you seriousely in the future
>

I downloaded the latest DirectX SDK to use OpenCV with Windows.
But no where it contains DirectShow  !!!
How to proceed..... I am quite naive in using it !!!
Please help as soon as possible !!!

_________________________________________________________________
Get up-to-date with movies, music and TV. Its happening on MSN Entertainment
http://content.msn.co.in/Entertainment/Default

Hi,

how about the speed-up version of SIFT, known as: SURF?
http://www.vision.ee.ethz.ch/~surf/


--- In OpenCV@yahoogroups.com, "darioroto" <darioroto@...> wrote:
>
> I'll try to be more precise:
>
> In our tests, SIFT extraction was too slow for real time, so as a
> result the tracker gets lost as we can't process enough frames per
> second.
> On the other hand, Harris corner extractor is fast enough (we've
> reached 15 fps in good conditions, with acceptable navigation, but
> not stable enough), but most
> of the Harris corners are features that are not stable over time,
as
> lots of them are light reflexions and stuff like this, so when the
> robot moves he doesn't find the same features again and it's hard
to
> get stability.
> Some people suggests us to use artags, but the goal was to navigate
> in an unknown environment...
> thanks
>
> darío
>
> --- In OpenCV@yahoogroups.com, "Peaceful Monkey"
> <anthony.remazeilles@> wrote:
> >
> > Could you be precise when you say that the harris corner are not
> stable
> > for navigation ? Are you talking about the feature tracker or the
> > feature extractor ?   Cordelia Schmid
> > <http://lear.inrialpes.fr/people/schmid/publications.html>  did a
> really
> > good job showing the property of different feature extractor. KLT
> > tracker has been widely use within so many applications....
> >
> > Extraction and tracking (and even matching) are not the same
thing.
> It
> > is not clear to understand which part you would like to improve
in
> your
> > application
> >
> > Thanks
> >
> > Anthony
> >
> > --- In OpenCV@yahoogroups.com, "darioroto" <darioroto@> wrote:
> > >
> > > hi,
> > >
> > > we're a group that is working on a humanoid robot, and we're
> trying to
> > > implement a visual SLAM program to navigate in an unknown
> environment.
> > >
> > > Has anybody tried to implement SIFT landmark detection in
openCV?
> > > We've tried with Harris corners and also with eigenvalues (both
> from
> > > cvGoodFeaturesToTrack) but they're not stable enough to
navigate.
> Has
> > > anybody tried SIFT landmarks for navigation? We haven't
> implemented it
> > > yet, but the few examples that we've found on the net where
> > reeeeeeally
> > > slow, so we can't try them, as we need something to navigate in
> real
> > > time...
> > >
> >
> >
> >
> > [Non-text portions of this message have been removed]
> >
>

HI,
Happy new year:)

I just installed opencv-1.0.0 in my Mandriva2007. I followed the
instruction on
http://opencvlibrary.sourceforge.net/CompileOpenCVUsingLinux
to compile sample C codes in directory opencv-1.0.0/samples/c/ and get
the corresponding excutable files. While when I run these excutable
files, it gave error:

OpenCV ERROR: Unspecified error (The function is not implemented.
Rebuild the library with Windows, GTK+ 2.x or Carbon support)
in function cvNamedWindow, window.cpp(71)
Terminating the application...
called from cvUnregisterType, cxpersistence.cpp(4933)
Terminating the application...
called from cvUnregisterType, cxpersistence.cpp(4933)
Terminating the application...
called from cvUnregisterType, cxpersistence.cpp(4933)
Terminating the application...
called from cvUnregisterType, cxpersistence.cpp(4933)
Terminating the application...
called from cvUnregisterType, cxpersistence.cpp(4933)
Terminating the application...
called from cvUnregisterType, cxpersistence.cpp(4933)
Terminating the application...
called from cvUnregisterType, cxpersistence.cpp(4933)
Terminating the application...
called from cvUnregisterType, cxpersistence.cpp(4933)
Terminating the application...
called from cvUnregisterType, cxpersistence.cpp(4933)
Terminating the application...


How can I fix this problem?
I am new in linux. If could, please give the details of the solution.
THanks in advance:)

Hi everybody, I'm  posting this again since nobody answered my last
post, hope I get lucky this time.

Im new to the group, and would like some points, on how to develope a
shape recognition machine for bottles.
A machine will have to detect between 10 different bottle shapes, and
tell what brand they are. All 10 bottles have different shapes
I just want some options from where to start looking.

Thanks all,

Best regards,
Alejandro

Are all views from the side with the bottles standing up?
Are all the images of single bottles only?
Are all the bottles symmetric about the vertical axis?

If yes to all 3 questions, then you should threshold to
obtain a binary image with a single blob. Then use the
blobslib to determine a set of features. Then classify
according to these features: (area, perimeter, length,
width, moments)

- Dave Grossman

________________________________

From: OpenCV@yahoogroups.com [mailto:OpenCV@yahoogroups.com] On Behalf Of
alekatz
Sent: Wednesday, January 03, 2007 9:20 AM
To: OpenCV@yahoogroups.com
Subject: [OpenCV] BOTTLE SHAPE RECOGNITION MACHINE QUESTION




Hi everybody, I'm posting this again since nobody answered my last
post, hope I get lucky this time.

Im new to the group, and would like some points, on how to develope a
shape recognition machine for bottles.
A machine will have to detect between 10 different bottle shapes, and
tell what brand they are. All 10 bottles have different shapes
I just want some options from where to start looking.

Thanks all,

Best regards,
Alejandro

Thanks Dave,
All the questions have a YES

Really greatfull with your suggestions,

Greetings
Alejandro Katz


--- Dave Grossman <dgrossman2@...> wrote:

> Are all views from the side with the bottles
> standing up?
> Are all the images of single bottles only?
> Are all the bottles symmetric about the vertical
> axis?
>
> If yes to all 3 questions, then you should threshold
> to
> obtain a binary image with a single blob. Then use
> the
> blobslib to determine a set of features. Then
> classify
> according to these features: (area, perimeter,
> length,
> width, moments)
>
> - Dave Grossman
>
> ________________________________
>
> From: OpenCV@yahoogroups.com
> [mailto:OpenCV@yahoogroups.com] On Behalf Of
> alekatz
> Sent: Wednesday, January 03, 2007 9:20 AM
> To: OpenCV@yahoogroups.com
> Subject: [OpenCV] BOTTLE SHAPE RECOGNITION MACHINE
> QUESTION
>
>
>
>
> Hi everybody, I'm posting this again since nobody
> answered my last
> post, hope I get lucky this time.
>
> Im new to the group, and would like some points, on
> how to develope a
> shape recognition machine for bottles.
> A machine will have to detect between 10 different
> bottle shapes, and
> tell what brand they are. All 10 bottles have
> different shapes
> I just want some options from where to start
> looking.
>
> Thanks all,
>
> Best regards,
> Alejandro
>
>
>
>


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

Hi all,
I'm trying to implement the camshiftdemo sample with a little
modification.
Instead of selecting the initial search area with mouse, I'm selecting
it using some other method.
I can select the initial tracking window successfully as seen here:
http://img412.imageshack.us/img412/9256/testimgjq9.jpg
But after I apply the cvCamShift function, "track_box" seems to have
no size. When I use cvEllipseBox to draw it, it appears as a dot at
the center of the initial tracking window and it never moves or
changes it's size.
Code is a little long so I'm trying to paste the code as a summary:

...
color_img = cvQueryFrame(input_video);
cvCvtColor(color_img, hsv_img, CV_BGR2HSV);
...
cvSplit(hsv_img, hue, 0, 0, 0 );
track_window = cvBoundingRect(contour, 0); //this is the rectangle
shown in the picture
cvSetImageROI(hue, track_window);
cvSetImageROI(mask, track_window);
...
int bin_w;
cvCalcHist( &hue, hist, 0, mask );
cvGetMinMaxHistValue( hist, 0, &max_val, 0, 0 );
cvConvertScale( hist->bins, hist->bins, max_val ? 255. / max_val : 0.,
0 );
cvResetImageROI( hue );
cvResetImageROI( mask );
cvZero( histimg );
bin_w = histimg->width / hdims;
//image that shows the histogram of the tracked window
for(int i = 0; i < hdims; i++ )
{
   int val = cvRound( cvGetReal1D(hist->bins,i)*histimg->height/255 );
   CvScalar color = hsv2rgb(i*180.f/hdims);
   cvRectangle( histimg, cvPoint(i*bin_w, histimg->height),
cvPoint((i+1)* bin_w, histimg->height - val),
       color, -1, 8, 0 );
}

cvCalcBackProject(&hue, backproject, hist);
cvAnd( backproject, mask, backproject, 0 );
cvCamShift( backproject, track_window, cvTermCriteria( CV_TERMCRIT_EPS
| CV_TERMCRIT_ITER, 10, 1 ),
       &track_comp, &track_box );
track_window = track_comp.rect;
//I'm not using "if( image->origin ) track_box.angle = -track_box.angle;"
//part, may that be the problem?

cvEllipseBox( camshift_img, track_box, CV_RGB(255,0,0), 3, CV_AA, 0 );
cvShowImage("CamShift", camshift_img);
for(nFrame; nFrame < nFrameCount; nFrame++ )
{
   color_img = cvQueryFrame(input_video);
...
   cvCvtColor( color_img, hsv_img, CV_BGR2HSV );
   cvSplit( hsv_img, hue, 0, 0, 0 );
   cvCalcBackProject(&hue, backproject, hist);
   cvAnd( backproject, mask, backproject, 0 );
   cvCamShift( backproject, track_window, cvTermCriteria(
CV_TERMCRIT_EPS | CV_TERMCRIT_ITER, 10, 1 ),
         &track_comp, &track_box );
   track_window = track_comp.rect;
   track_box.angle = -track_box.angle;
   camshift_img = cvCloneImage(color_img);
   cvEllipseBox( camshift_img, track_box, CV_RGB(0,0,255), 3, CV_AA, 0 );
...
}

Why does not "track_box" have a size?
This is a little urgent so I would really appreciate any help.

Hello,

Is there any way to use cvcam to capture from a webcam without having
any window rendering the stream? (Other than cvDestroyWindow just after
launch...)

I've seen that cvcam is really faster than highgui, using directshow,
but I'd like to use it the same way highgui does (asking for frames, not
with a callback). I'm thinking about make a callaback that just cvClone
to other image, and then manage it for myself...

Any advice or clue will be very appreciated, I'm really close to a
project deadline.
Thanks in advance,
Alberto Jaspe.

If I don't use the track_box by using:

cvCamShift( backproject, track_window, cvTermCriteria( CV_TERMCRIT_EPS
| CV_TERMCRIT_ITER, 10, 1 ), &track_comp);
instead of
cvCamShift( backproject, track_window, cvTermCriteria( CV_TERMCRIT_EPS
| CV_TERMCRIT_ITER, 10, 1 ), &track_comp, &track_box );

cvRectangle(camshift_img, cvPoint(track_window.x, track_window.y),
   cvPoint(track_window.x+track_window.width,
   track_window.y+track_window.height), CV_RGB(255,0,0), 1, 4, 0);
instead of
cvEllipseBox( camshift_img, track_box, CV_RGB(255,0,0), 3, CV_AA, 0 );

I see the initial window drawn as the image I provided in my previous
email, and that rectangle stays the same during the whole video.
What am I missing?

--- In OpenCV@yahoogroups.com, "Eren Aykin" <erenaykin@...> wrote:
>
> Hi all,
> I'm trying to implement the camshiftdemo sample with a little
> modification.
> Instead of selecting the initial search area with mouse, I'm selecting
> it using some other method.
> I can select the initial tracking window successfully as seen here:
> http://img412.imageshack.us/img412/9256/testimgjq9.jpg
> But after I apply the cvCamShift function, "track_box" seems to have
> no size. When I use cvEllipseBox to draw it, it appears as a dot at
> the center of the initial tracking window and it never moves or
> changes it's size.
> Code is a little long so I'm trying to paste the code as a summary:
>
> ...
> color_img = cvQueryFrame(input_video);
> cvCvtColor(color_img, hsv_img, CV_BGR2HSV);
> ...
> cvSplit(hsv_img, hue, 0, 0, 0 );
> track_window = cvBoundingRect(contour, 0); //this is the rectangle
> shown in the picture
> cvSetImageROI(hue, track_window);
> cvSetImageROI(mask, track_window);
> ...
> int bin_w;
> cvCalcHist( &hue, hist, 0, mask );
> cvGetMinMaxHistValue( hist, 0, &max_val, 0, 0 );
> cvConvertScale( hist->bins, hist->bins, max_val ? 255. / max_val : 0.,
> 0 );
> cvResetImageROI( hue );
> cvResetImageROI( mask );
> cvZero( histimg );
> bin_w = histimg->width / hdims;
> //image that shows the histogram of the tracked window
> for(int i = 0; i < hdims; i++ )
> {
>   int val = cvRound( cvGetReal1D(hist->bins,i)*histimg->height/255 );
>   CvScalar color = hsv2rgb(i*180.f/hdims);
>   cvRectangle( histimg, cvPoint(i*bin_w, histimg->height),
> cvPoint((i+1)* bin_w, histimg->height - val),
>       color, -1, 8, 0 );
> }
>
> cvCalcBackProject(&hue, backproject, hist);
> cvAnd( backproject, mask, backproject, 0 );
> cvCamShift( backproject, track_window, cvTermCriteria( CV_TERMCRIT_EPS
> | CV_TERMCRIT_ITER, 10, 1 ),
>       &track_comp, &track_box );
> track_window = track_comp.rect;
> //I'm not using "if( image->origin ) track_box.angle =
-track_box.angle;"
> //part, may that be the problem?
>
> cvEllipseBox( camshift_img, track_box, CV_RGB(255,0,0), 3, CV_AA, 0 );
> cvShowImage("CamShift", camshift_img);
> for(nFrame; nFrame < nFrameCount; nFrame++ )
> {
>   color_img = cvQueryFrame(input_video);
> ...
>   cvCvtColor( color_img, hsv_img, CV_BGR2HSV );
>   cvSplit( hsv_img, hue, 0, 0, 0 );
>   cvCalcBackProject(&hue, backproject, hist);
>   cvAnd( backproject, mask, backproject, 0 );
>   cvCamShift( backproject, track_window, cvTermCriteria(
> CV_TERMCRIT_EPS | CV_TERMCRIT_ITER, 10, 1 ),
>         &track_comp, &track_box );
>   track_window = track_comp.rect;
>   track_box.angle = -track_box.angle;
>   camshift_img = cvCloneImage(color_img);
>   cvEllipseBox( camshift_img, track_box, CV_RGB(0,0,255), 3, CV_AA,
0 );
> ...
> }
>
> Why does not "track_box" have a size?
> This is a little urgent so I would really appreciate any help.
>

Hello Tebish,

DirectShow has been translated from DirectX to the MS Platform SDK.
There you can find it.

Tabish Khan escribió:
>
> I downloaded the latest DirectX SDK to use OpenCV with Windows.
> But no where it contains DirectShow !!!
> How to proceed..... I am quite naive in using it !!!
> Please help as soon as possible !!!
>

Thanks a lot !!!
We need to create a gesture recognition or interpretation
application.....there are few samples in OpenCv too which never run...it
says "some" file is missing...we need on work on that too...Please do
suggest how to proceed !!!


>From: Alberto Jaspe <ajaspe@...>
>Reply-To: OpenCV@yahoogroups.com
>To: OpenCV@yahoogroups.com
>Subject: Re: [OpenCV] DirectShow
>Date: Wed, 03 Jan 2007 20:22:19 +0100
>
>Hello Tebish,
>
>DirectShow has been translated from DirectX to the MS Platform SDK.
>There you can find it.
>
>Tabish Khan escribió:
> >
> > I downloaded the latest DirectX SDK to use OpenCV with Windows.
> > But no where it contains DirectShow !!!
> > How to proceed..... I am quite naive in using it !!!
> > Please help as soon as possible !!!
> >
>
>
>

_________________________________________________________________
Mega Fare sale on domestic airlines. Call 1-800-11-8747 now!
http://india.makemytrip.com/?Camp=MSN&attrib=Dec/Tag/Mega

I write these code:
/////////////////////////////////////
float a[]={1,2,3};
CvMat ma=cvMat(3,1,CV_32FC1,a);
CvMat *mb = cvCreateMat(4,1,CV_32FC1);
cvConvertPointsHomogenious(&ma,mb);
//////////////////////////////////////

but it return error:
"Either the number of channels or columns or rows must be =1"

I am confused.
What'r wrong?

Thank you.

Hi there,

I'd like to know if anyone has tried to derive 2 projection matrices
(for each one of the images) from a fundamental matrix in OpenCV style.
I'm trying to improve a 3D reconstruction algorithm, by using a
Levenberg-Marquardt optimization, and it woul'd be very helpfull.

Best compliments,
Rafael Bastos

hai
       I am planning to develop a project related to 3D face modeling from
images. Input will be from two images(Frontal and Profile view of face).I want
to discuss some issues and if anyone done such things plz plz reply me


  __________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

[Non-text portions of this message have been removed]

hai
     I am planning to develop a project related to 3D face modeling
from images. Input will be from two images(Frontal and Profile view of
face).I want to discuss some issues and if anyone done such things plz
plz reply me

hi everyone,

my problems not related with openCV, but since you all are doing image
processing, i thought of asking this question.
is there any free image database available for handwritten english
characters,digits etc.(so i can download the images)

thank you.
randeel

  __________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

[Non-text portions of this message have been removed]

Hi,
One question regarding the definition of Haar-like features. I want to
add to the "features database" other features
(http://www.hpl.hp.com/research/isl/redeye/redeye-icpr.pdf    - page 3).
Where in the code can I do it? Where are the currently features defined?
I am using OpenCV_1.0. Thanks for responses. Tomas

hi all,
i need some code on sobel edge detection in c++ on a upload image can
someone help  me

Dear Erik:

       I have one VX-6000. It works very well under Windows XP with openCV.

Sincerely yours,
shunza

   ----- Original Message -----
   From: erik_cuevas2003
   To: OpenCV@yahoogroups.com
   Sent: Friday, December 22, 2006 7:56 PM
   Subject: [OpenCV] Microsoft Lifecam VX-6000


   Hello,

   I want to buy the Microsoft Lifecam VX-6000 with
   1200x1000 video and 5MP photo.

   My question is, whether the camera would tested under
   Windows XP with openCV. Can someone help me?

   thx,
   Erik.




[Non-text portions of this message have been removed]

hi all,
i am using visual c++ to ipload a image and find location of the
license plate. but iam having problems with the sobel edge detection.
can some one send me a code where an image can be uplaoded and sobel
applied

Hi,

I am new in c++ programming and not experienced in image processing
and I will use OpenCV in my master thesis. I will work on video
content analysis. My starting point will be "shot boundary detection"

I have installed OpenCV and made some progress on working with
images. I am reading the openCV tutorial by laganier at:
http://www.site.uottawa.ca/~laganier/tutorial/opencv+directshow/cvisio
n.htm

I will work on the video files (mpeg 1 files) not with the camera.
This tutorial is from 2004. It talks about the DirectX sdk and
DirectShow capabilities for video processing but it seems to me a bit
complex.

First of all, if you know some video processing tutorials they will
help me lot.

Secondly, should i use DirectShow? is it necessary? I already waste
alot of time finding the direct show components because they are no
more included in the DirectX sdk nor Platform SDK. And they are
difficult to understand for me.

How can i open a mpeg1 file using openCV?

Thanks alot.

I have openCV running in xcode as frameworks by combining the directions
from the following 2 sites:

This first site is for version 0.97 of opencv, but its still pretty relevant
for version 1.0 as well.  Additionally, I installed jasper-1.7 and
open-exr1.4.  I also installed ffmpeg, but the new version 1.0 can use
Quicktime libs as well.

http://www.christophseibert.de/weblog/en/it/computer-vision/tools/opencv-mac.htm\
l

After following the above directions, you should be able to get the command
line version of opencv 1.0 working.  I was able to run the facedetect
examples on the Mac and other examples no problem.  In addition, the make
check command passed all tests.  One thing to note here, is that if you are
using fink, all your libs are in /sw/lib instead of /usr/local/lib, so you
will have to change the LDFLAGS settings as described in the above link.

Ok, now to get opencv working in xcode, I built frameworks using the
directions and build environment found here.  I used the beta version for my
build:

http://www.lestang.org/article45.html

I essentially added the entire opencv release into the appropriate directory
of the build environment found above.  I added include paths and lib paths
to appropriate libraries that opencv needs to run.  All of my libs are in
the fink directory /sw/lib.  Once the frameworks build properly, I copied
those to /Library/Frameworks.  All of my subsequent xcode projects (latest
being a cocoa application) simply includes the cv.framefork, cvaux.framework,
cxcore.framework, and highgui.framework.

Currently I can call any opencv functions from cocoa applications no
problem.  I can actually run the facedetect example, and it brings up
windows to show the image with bounding box as well.  It totally works
perfectly.  You will want to use the -bind-at-load linker directive in all
of your projects to get this to work appropriately.

I do have the complete build environment that is working for me.  I just
tarred it up and will place at http://www.cs.northwestern.edu/~jskitz/opencv.
Enjoy!

Hope the above helps a little.

Happy New Year,

Jason

On 12/29/06, phoebehbright <phoebehbright@...> wrote:
>
>   After a couple of days of googling and experimentation, I can now
> create c++ projects and run them in XCode and online. However, I am
> still at a loss to know how to use OpenCV. I enclose a few links in
> case they help anyone else.
>
> No clues anywhere on the net on how to include opencv in XCode (I know
> it's obvious once you know!). This might help others:
>
http://www.d.umn.edu/~cprince/PubRes/IKAROS/OpenCV/install.html<http://www.d.umn\
.edu/%7Ecprince/PubRes/IKAROS/OpenCV/install.html>
>
> And this is almost it but not quite: http://lestang.org/article45.html
>
> More opencv links here: http://dave.thehorners.com/content/view/95/67/
>
> Tried running from the command line as suggested here:
> http://dave.thehorners.com/content/view/95/67/
>
> but get this:
>
> g++ hello.cpp -o hello -I /usr/local/include/opencv -L
> /usr/local/lib -lm -lcv -lhighgui -lcvaux
> /usr/bin/ld: -L: directory name missing
> collect2: ld returned 1 exit status
>
> Looked at the options for g++ in OSX and guess -L is to do with
> linking but no info provided.
>
> Am running out of steam here. What I would dearly love, is a zipped
> up XCode project that included all the openCV stuff and a hello world
> example that I can click build and run and it does! Huge goodwill to
> anyone able to provide this!!
>
> Happy New Year.
>
>
>


[Non-text portions of this message have been removed]