OpenVIDIA : Parallel GPU Computer Vision

Tracker

 - Bugs ( 0 open / 0 total )
Bug Tracking System

 - Support Requests ( 1 open / 1 total )
Tech Support Tracking System

 - Patches ( 1 open / 1 total )
Patch Tracking System

 - Feature Requests ( 0 open / 1 total )
Feature Request Tracking System

 Forums

 Doc Manager

 Mailing Lists

 Screenshots

 Task Manager

 CVS Tree

 Released Files

Related Projects

Comparametric Toolkit

Reference Links

Open GL Reference

nVIDIA cg homepage

GeForce FX Overclocking

GPGPU

gpu's arent just good for vision and computation - some clever folks have started using them for games as well. ;) har har linux-games.net

7 Years of Graphics Card History

Page design based on Blosxom (which was used originally before the move to sourceforge).

This page best viewed with Dillo, Lynx, w3m, Mozilla-Firefox, Galeon, Epiphany, konqueror . . .

Programming with libopenvidia, version 0.8 and up

The following show how to use the current functions included in libopenvidia 0.8.0. Work is underway to include all the features of 0.07 into 0.8.

The following examples are snippets from the example programs which can be found in /examples

Click here for class documentation

Additionally, note some objects are subclassed CommonC++ objects, and others from STL vectors and pairs.

Firewire Image Capture

  //construct a 1394 capture object, using 320x240 resolution.
  // options are 640x480, 320x240, 160x120
  Dc1394 CamSource(320,240);

  // make a texture
  glGenTextures(1, &tex);              // texture 
  glTexEnvi( GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE,  GL_REPLACE );
  glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex);
  glTexImage2D(GL_TEXTURE_RECTANGLE_NV, 0, GL_RGBA, width,height, 0,
                GL_RGB, GL_UNSIGNED_BYTE,NULL );
  
  //start the camera capture
  CamSource.start();

   // rendering loop
   while(1) {

        //wait for a new frame to be captured.
        // this can be removed if you dont mind re-using a previous frame.
        CamSource.wait();
        //lock the image data so it is not updated while we are capturing.
        CamSource.lock();

        glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex);
        glTexSubImage2D(GL_TEXTURE_RECTANGLE_NV, 0, 0,0, width,height,
                         GL_RGB, GL_UNSIGNED_BYTE,CamSource.ptr());


        //free the image data so it can be updated again.
        CamSource.unlock();

        //use the image...
   }
 

Generic CG image processing using FBO_Filter

            // Generate a texture to hold the input image 
            glGenTextures(1, &tex);              // texture 
            glTexEnvi( GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE,  GL_REPLACE );
            glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex);
            glTexImage2D(GL_TEXTURE_RECTANGLE_NV, 0, GL_RGBA, width,height, 0,
                         GL_RGB, GL_UNSIGNED_BYTE, Im->rgb_data );

            // Generate a texture to hold the output image.  Set the output texture's 
            // internal format to the desired output precision 
            glGenTextures(1, &oTex);              // texture 
            glBindTexture(GL_TEXTURE_RECTANGLE_NV, oTex);
            glTexImage2D(GL_TEXTURE_RECTANGLE_NV, 0, GL_RGBA, width,height, 0,
                         GL_RGB, GL_UNSIGNED_BYTE, NULL );

            // Create the filter.
            filter = new FBO_Filter(CG_PROFILE_FP30,argv[1], oTex, width, height);

            ...

            // rendering loop 
            while(true) {

                 //apply the filter.  fills in oTex 
                 filter->apply( tex, false);

                 //show the resulting texture oTex 
                 glEnable(GL_TEXTURE_RECTANGLE_NV);
                 glBindTexture(GL_TEXTURE_RECTANGLE_NV, oTex) ;

                 // draw the 'tex' texture containing the framebuffer rendered drawing.
                 glBegin(GL_QUADS);
                    glTexCoord2f(0, height);
                    glVertex3f( 0.0, 1.0, d );

                    glTexCoord2f(0, 0);
                    glVertex3f( 0.0, 0.0, d );

                    glTexCoord2f(width, 0);
                    glVertex3f( 1.0, 0.0, d );

                    glTexCoord2f(width,height);
                    glVertex3f( 1.0, 1.0, d );
                glEnd();
            }

Detecting and extracting features in an image

Finds features in an image and returns unique vectors for each. Implemnetation is based upon SIFT keys.

Typical usage:

   //create the feature tracking object. give it the size of the images it will operate on.
   ft = new featureTrack(sourcewidth, sourceheight );

   // make a texture
   glGenTextures(1, &tex);              // texture 
   glTexEnvi( GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE,  GL_REPLACE );
   glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex);
   glTexImage2D(GL_TEXTURE_RECTANGLE_NV, 0, GL_RGBA, sourcewidth,
                sourceheight, 0, GL_RGB, GL_UNSIGNED_BYTE,NULL );

  //rendering loop
  while(1) {
    //update the image frame...
    glBindTexture(GL_TEXTURE_RECTANGLE_NV, tex );
    glTexSubImage2D(GL_TEXTURE_RECTANGLE_NV, 0, 0,0, sourcewidth, sourceheight,
                    GL_RGB, GL_UNSIGNED_BYTE, sourcedata );

    // retrieve features from the image.

    Scene *s = ft->getScene(tex);

    // scene is a STL vector of features.
    cerr<<"Scene with "<< s->features.size() <<" features.\n";

    //operate on the scene, s

    ...

    // toss the scene away after we're done to avoid memory leak
     delete(s);
   }