Structured forests for fast edge detection

Introduction

In this tutorial you will learn how to use structured forests for the purpose of edge detection in an image.

Examples

image

image

image

image

image

image

image

image

image

image

image

image

Note

binarization techniques like Canny edge detector are applicable to edges produced by both algorithms (Sobel and StructuredEdgeDetection::detectEdges).

Source Code

/**************************************************************************************
The structered edge demo requires you to provide a model.
This model can be found at the opencv_extra repository on Github on the following link:
https://github.com/opencv/opencv_extra/blob/master/testdata/cv/ximgproc/model.yml.gz
***************************************************************************************/

#include <opencv2/ximgproc.hpp>
#include "opencv2/highgui.hpp"
#include "opencv2/core/utility.hpp"
#include <iostream>

using namespace cv;
using namespace cv::ximgproc;

const char* keys =
{
    "{i || input image name}"
    "{m || model name}"
    "{o || output image name}"
};

int main( int argc, const char** argv )
{
    bool printHelp = ( argc == 1 );
    printHelp = printHelp || ( argc == 2 && String(argv[1]) == "--help" );
    printHelp = printHelp || ( argc == 2 && String(argv[1]) == "-h" );

    if ( printHelp )
    {
        std::cout << "\nThis sample demonstrates structured forests for fast edge detection\n"
               "Call:\n"
               "    structured_edge_detection -i=in_image_name -m=model_name [-o=out_image_name]\n\n";
        return 0;
    }

    CommandLineParser parser(argc, argv, keys);
    if ( !parser.check() )
    {
        parser.printErrors();
        return -1;
    }

    String modelFilename = parser.get<String>("m");
    String inFilename = parser.get<String>("i");
    String outFilename = parser.get<String>("o");

    Mat image = imread(inFilename, 1);
    if ( image.empty() )
        CV_Error(Error::StsError, String("Cannot read image file: ") + inFilename);

    if ( modelFilename.size() == 0)
        CV_Error(Error::StsError, String("Empty model name"));

    image.convertTo(image, DataType<float>::type, 1/255.0);

    Mat edges(image.size(), image.type());

    Ptr<StructuredEdgeDetection> pDollar =
        createStructuredEdgeDetection(modelFilename);
    pDollar->detectEdges(image, edges);

    // computes orientation from edge map
    Mat orientation_map;
    pDollar->computeOrientation(edges, orientation_map);

    // suppress edges
    Mat edge_nms;
    pDollar->edgesNms(edges, orientation_map, edge_nms, 2, 0, 1, true);

    if ( outFilename.size() == 0 )
    {
        namedWindow("edges", 1);
        imshow("edges", edges);
        namedWindow("edges nms", 1);
        imshow("edges nms", edge_nms);
        waitKey(0);
    }
    else
        imwrite(outFilename, 255*edges);

    return 0;
}

Explanation

1. Load source color image

   cv::Mat image = cv::imread(inFilename, 1);
   if ( image.empty() )
   {
       printf("Cannot read image file: %s\n", inFilename.c_str());
       return -1;
   }

2. Convert source image to [0;1] range

   image.convertTo(image, cv::DataType<float>::type, 1/255.0);

3. Run main algorithm

   cv::Mat edges(image.size(), image.type());
   
   cv::Ptr<StructuredEdgeDetection> pDollar =
       cv::createStructuredEdgeDetection(modelFilename);
   pDollar->detectEdges(image, edges);

4. Show results

   if ( outFilename == "" )
   {
       cv::namedWindow("edges", 1);
       cv::imshow("edges", edges);
   
       cv::waitKey(0);
   }
   else
       cv::imwrite(outFilename, 255*edges);

Literature

For more information, refer to the following papers : [50] [127]