Tutorial "Image -> ImageAdapter: some simple image manipulations"

Table of Contents

• Preliminary
• Basic Image manipulations
  • Image cropping
  • 2D slice image extraction from 3D volume
  • 2D Image extraction from 3D volume
  • Create image with color Functor

Author(s) of this documentation:

Bertrand Kerautret

Preliminary

The GenericReader and GenericWriter class allow you to import/export different image formats depending of the optional DGtal dependancies (ITK).

Todo

Exercise Construct a simple file format converter starting from this file template:

 
#include "DGtal/base/Common.h"
#include "DGtal/images/ImageContainerBySTLVector.h"
#include "DGtal/io/readers/GenericReader.h"
#include "DGtal/io/writers/GenericWriter.h"
#include "DGtal/helpers/StdDefs.h"
 
#include <boost/program_options/options_description.hpp>
#include <boost/program_options/parsers.hpp>
#include <boost/program_options/variables_map.hpp>
 
 
using namespace std;
using namespace DGtal;
using namespace Z3i;
 
namespace po = boost::program_options;
 
 
int main( int argc, char** argv )
{
  
  typedef ImageContainerBySTLVector<Z2i::Domain, unsigned char>  Image2D;
  
  // parse command line ----------------------------------------------
  po::options_description general_opt ( "Allowed options are: " );
  general_opt.add_options()
    ( "help,h", "display this message." )
    ( "input,i", po::value<std::string>(), "Input vol file." )
    ( "output,o", po::value<string>(),"Output OFF filename." );
 
 
  bool parseOK=true;
  po::variables_map vm;
  try{
    po::store(po::parse_command_line(argc, argv, general_opt), vm);
  }catch(const std::exception& ex){
    parseOK=false;
    trace.info()<< "Error checking program options: "<< ex.what()<< endl;
  }
  po::notify ( vm );
  if (!parseOK || vm.count ( "help" ) ||argc<=1 || !vm.count("input")||! vm.count("output"))
    {
      trace.info() << "exercise ImageIO."<<std::endl
                   << std::endl << "Basic usage: "<<std::endl
                   << "\tvol2off --input <volFileName> --o <OffOutputFileName> "<<std::endl
                   << general_opt << "\n";
      return 0;
    }
 
 
 
  return 0;
}

Complements:

• Test the same reading/export on 3D volume (vol pgm3D format).

Basic Image manipulations

The aim of this tutorial is to propose some basic image manipulations in particular by exploiting the image adapter classes (ImageAdapter) with different functors.

Image cropping

The ImageAdapter and ConstImageAdapter can be very efficient to define for instance a cropped image (see documentation given in the Images module). To persform a simple crop you can apply the following steps:

• Load the source image to be cropped.
• Define type associated to the ConstImageAdapter.
• Define a sub domain and construct the cropped image.
• Export the result.

If you use image of DGtal/examples/samples/church.pgm with the given domain, you should obtain the following image:

Image cropped to the domain defined by points (170,466) and (316, 745).

Complements:

• Add options to define the input/output files and region bounds.
• Apply the same cropping to 3D images.

2D slice image extraction from 3D volume

In the same spirit than previous example, you can also extract a 2D slice image from a 3D volume image.

To apply the extraction you can follow these steps:

• Construct a ConstImageAdapter with the following types:
  • type of the source Image (Image3D).
  • the domain type of the resulting image (Image2D::Domain).
  • the functor taking in input 2D data and return 3D points ( DGtal::functors::Projector<Z3i::Space>) to access to source image data.
  • the type of value of the 3D image (Image3D::Value)
  • the functor type to be applied on the source image value during the reading process.

    typedef ConstImageAdapter<Image3D, Image2D::Domain, functors::Projector<Z3i::Space>,
                               Image3D::Value,  functors::Identity >  SliceImageAdapter;

• Construct the domain of the resulting image.
• Construct the functor to embed the 2D point into the 3D volume (by using the Projector class returning 3D points from 2D coordinates).

If you have used the vol file: DGtal/examples/samples/lobster.vol and a slice number, you should obtain the following image:

Result of slice image number 20 on Z axis.

Todo

Complementary exercise:

• Apply the same slice extraction on axis X and Y intersecting the center point of the 3D Image.

Hints:

• To construct the domain of the resulting image you can use a projector:

  DGtal::functors::Projector<Z2i::Space >  proj(2);
  DGtal::Z2i::Domain domain2D(proj(image3d.domain().lowerBound()),
                              proj(image3d.domain().upperBound()));

2D Image extraction from 3D volume

The functor Point2DEmbedderIn3D defined in the BasicPointFunctors class allows to embed a point of an 2D domain into a 3D one. By this way you can extract a 2D image from a 3D volume just by using this functor as pointFunctor in the ConstImageAdapter.

To construct a Point2DEmbedderIn3D you have to give the domain of the 3d image, the origin point with an upper point on axis1 and on axis2.

Todo

Exercise: Construct a program able to extract an 2D image from a given volume and 3 3D points. For instance you can to extract the 2D image defined by the points:

• Origin (120, 150, 10)
• Upper point1 (220, 220, 10)
• Upper point2 (120, 150, 50) You should obtain the following 2D image:

Result of slice image number 20 on Z axis.

Create image with color Functor

By using the HueShadeColorMap as functor in the exportPPM method (of PPMWriter), transform a grayscale image into color. For instance with the previous image you should obtain:

Color image from HueShadeColorMap functor.