Tutorial: Automatic thresholding

Table of Contents

• Introduction
• Example code
• Next tutorial

Introduction

This tutorial will show you how to automatically threshold, binarize an image using different methods:

• Huang [21]
• Intermodes [40]
• Isodata [41]
• Mean [17]
• Otsu [38]
• Triangle [48]

These functions have been ported from the Auto Threshold ImageJ plugin and you can refer to the corresponding documentation for more information.

Example code

The following example also available in tutorial-autothreshold.cpp will demonstrate on a sample image the result of each of these methods:

 
#include <cstdlib>
#include <iostream>
#include <visp3/core/vpImage.h>
#include <visp3/gui/vpDisplayGDI.h>
#include <visp3/gui/vpDisplayOpenCV.h>
#include <visp3/gui/vpDisplayX.h>
#include <visp3/io/vpImageIo.h>
 
#if defined(VISP_HAVE_MODULE_IMGPROC)
#include <visp3/imgproc/vpImgproc.h>
#endif
 
int main(int argc, const char **argv)
{
#if defined(VISP_HAVE_MODULE_IMGPROC) && (defined(VISP_HAVE_X11) || defined(VISP_HAVE_GDI) || defined(VISP_HAVE_OPENCV))
  std::string input_filename = "grid36-03.pgm";
 
  for (int i = 1; i < argc; i++) {
    if (std::string(argv[i]) == "--input" && i + 1 < argc) {
      input_filename = std::string(argv[i + 1]);
    } else if (std::string(argv[i]) == "--help" || std::string(argv[i]) == "-h") {
      std::cout << "Usage: " << argv[0] << " [--input <input image>] [--help]" << std::endl;
      return EXIT_SUCCESS;
    }
  }
 
  vpImage<unsigned char> I;
  vpImageIo::read(I, input_filename);
 
  vpImage<unsigned char> I_res(3 * I.getHeight(), 3 * I.getWidth());
  I_res.insert(I, vpImagePoint(I.getHeight(), I.getWidth()));
 
#ifdef VISP_HAVE_X11
  vpDisplayX d;
#elif defined(VISP_HAVE_GDI)
  vpDisplayGDI d;
#elif defined(VISP_HAVE_OPENCV)
  vpDisplayOpenCV d;
#endif
  d.setDownScalingFactor(vpDisplay::SCALE_2);
  d.init(I_res);
 
  vpImage<unsigned char> I_huang = I;
  vp::autoThreshold(I_huang, vp::AUTO_THRESHOLD_HUANG);
  I_res.insert(I_huang, vpImagePoint());
 
  vpImage<unsigned char> I_intermodes = I;
  vp::autoThreshold(I_intermodes, vp::AUTO_THRESHOLD_INTERMODES);
  I_res.insert(I_intermodes, vpImagePoint(0, I.getWidth()));
 
  vpImage<unsigned char> I_isodata = I;
  vp::autoThreshold(I_isodata, vp::AUTO_THRESHOLD_ISODATA);
  I_res.insert(I_isodata, vpImagePoint(0, 2 * I.getWidth()));
 
  vpImage<unsigned char> I_mean = I;
  vp::autoThreshold(I_mean, vp::AUTO_THRESHOLD_MEAN);
  I_res.insert(I_mean, vpImagePoint(I.getHeight(), 0));
 
  vpImage<unsigned char> I_otsu = I;
  vp::autoThreshold(I_otsu, vp::AUTO_THRESHOLD_OTSU);
  I_res.insert(I_otsu, vpImagePoint(I.getHeight(), 2 * I.getWidth()));
 
  vpImage<unsigned char> I_triangle = I;
  vp::autoThreshold(I_triangle, vp::AUTO_THRESHOLD_TRIANGLE);
  I_res.insert(I_triangle, vpImagePoint(2 * I.getHeight(), 0));
 
  vpDisplay::display(I_res);
 
  vpDisplay::displayText(I_res, 30, 20, "Huang", vpColor::red);
  vpDisplay::displayText(I_res, 30, 20 + I.getWidth(), "Intermodes", vpColor::red);
  vpDisplay::displayText(I_res, 30, 20 + 2 * I.getWidth(), "IsoData", vpColor::red);
  vpDisplay::displayText(I_res, 30 + I.getHeight(), 20, "Mean", vpColor::red);
  vpDisplay::displayText(I_res, 30 + I.getHeight(), 20 + I.getWidth(), "Original", vpColor::red);
  vpDisplay::displayText(I_res, 30 + I.getHeight(), 20 + 2 * I.getWidth(), "Otsu", vpColor::red);
  vpDisplay::displayText(I_res, 30 + 2 * I.getHeight(), 20, "Triangle", vpColor::red);
 
  vpDisplay::flush(I_res);
  vpDisplay::getClick(I_res);
 
  return EXIT_SUCCESS;
#else
  (void)argc;
  (void)argv;
  return 0;
#endif
}

These functions are provided in a vp:: namespace and accessible using this include:

#include <visp3/imgproc/vpImgproc.h>

The code to use is pretty straightword:

• Huang:

    vpImage<unsigned char> I_huang = I;
    vp::autoThreshold(I_huang, vp::AUTO_THRESHOLD_HUANG);

• Intermodes:

    vpImage<unsigned char> I_intermodes = I;
    vp::autoThreshold(I_intermodes, vp::AUTO_THRESHOLD_INTERMODES);

• IsoData:

    vpImage<unsigned char> I_isodata = I;
    vp::autoThreshold(I_isodata, vp::AUTO_THRESHOLD_ISODATA);

• Mean:

    vpImage<unsigned char> I_mean = I;
    vp::autoThreshold(I_mean, vp::AUTO_THRESHOLD_MEAN);

• Otsu:

    vpImage<unsigned char> I_otsu = I;
    vp::autoThreshold(I_otsu, vp::AUTO_THRESHOLD_OTSU);

• Triangle:

    vpImage<unsigned char> I_triangle = I;
    vp::autoThreshold(I_triangle, vp::AUTO_THRESHOLD_TRIANGLE);

The following image presents the results for each method:

Comparison of different binarizations using the threshold value returned by each method

The function vp::autoThreshold(vpImage<unsigned char> &, const vp::vpAutoThresholdMethod &, const unsigned char, const unsigned char) has two parameters to specify the pixel values to use for the background and the foreground. By default, it is (see vpImageTools::binarise(vpImage<Type> &, Type, Type, Type, Type, Type, const bool)):

Next tutorial

You can now read the Tutorial: Contours extraction from a binary image, to learn how to extract the contours from a binary image.