opencv实例matching_to_many_images不能使用surf解决

        这两天在试试opencv这个开源项目的实例，发现默认配置时matching_to_many_images居然无法执行，参数是SURF SURF FlannBased E:\opencv-test\2.4.6\opencv\samples\cpp\matching_to_many_images\query.png E:\opencv-test\2.4.6\opencv\samples\cpp\matching_to_many_images\train\trainImages.txt E:\opencv-test\2.4.6\opencv\samples\cpp\matching_to_many_images\result，实例的help中给出的。

        那就跟吧，初略跟进去，

        这两句的create没成功。

    featureDetector = FeatureDetector::create( detectorType );
    descriptorExtractor = DescriptorExtractor::create( descriptorType );

      再往里面细跟，发现FeatureDetector和DescriptorExtractor中居然没SURF这个类型的FeatureDetector，可是官方的资料中说支持SURF

      再看看features2d_init.cpp中的初始化，确实没有注册SURF。

     后面再细查资料才发现，原来老版本是有默认注册SURF到features2d，后来移到nonfree中了。只要对实例添加nonfree的初始化就可以使用了。

     改动如下：

     添加包含

     #include "opencv2/nonfree/nonfree.hpp"

     在代码开始时添加初始化

     cv::initModule_nonfree();

    修改后的代码如下：

#include "opencv2/highgui/highgui.hpp"
#include "opencv2/features2d/features2d.hpp"
#include "opencv2/contrib/contrib.hpp"
#include "opencv2/nonfree/nonfree.hpp"


#include <iostream>
#include <fstream>


using namespace cv;
using namespace std;


const string defaultDetectorType = "SURF";
const string defaultDescriptorType = "SURF";
const string defaultMatcherType = "FlannBased";
const string defaultQueryImageName = "../../opencv/samples/cpp/matching_to_many_images/query.png";
const string defaultFileWithTrainImages = "../../opencv/samples/cpp/matching_to_many_images/train/trainImages.txt";
const string defaultDirToSaveResImages = "../../opencv/samples/cpp/matching_to_many_images/results";


static void printPrompt( const string& applName )
{
    cout << "/*\n"
         << " * This is a sample on matching descriptors detected on one image to descriptors detected in image set.\n"
         << " * So we have one query image and several train images. For each keypoint descriptor of query image\n"
         << " * the one nearest train descriptor is found the entire collection of train images. To visualize the result\n"
         << " * of matching we save images, each of which combines query and train image with matches between them (if they exist).\n"
         << " * Match is drawn as line between corresponding points. Count of all matches is equel to count of\n"
         << " * query keypoints, so we have the same count of lines in all set of result images (but not for each result\n"
         << " * (train) image).\n"
         << " */\n" << endl;


    cout << endl << "Format:\n" << endl;
    cout << "./" << applName << " [detectorType] [descriptorType] [matcherType] [queryImage] [fileWithTrainImages] [dirToSaveResImages]" << endl;
    cout << endl;


    cout << "\nExample:" << endl
         << "./" << applName << " " << defaultDetectorType << " " << defaultDescriptorType << " " << defaultMatcherType << " "
         << defaultQueryImageName << " " << defaultFileWithTrainImages << " " << defaultDirToSaveResImages << endl;
}


static void maskMatchesByTrainImgIdx( const vector<DMatch>& matches, int trainImgIdx, vector<char>& mask )
{
    mask.resize( matches.size() );
    fill( mask.begin(), mask.end(), 0 );
    for( size_t i = 0; i < matches.size(); i++ )
    {
        if( matches[i].imgIdx == trainImgIdx )
            mask[i] = 1;
    }
}


static void readTrainFilenames( const string& filename, string& dirName, vector<string>& trainFilenames )
{
    trainFilenames.clear();


    ifstream file( filename.c_str() );
    if ( !file.is_open() )
        return;


    size_t pos = filename.rfind('\\');
    char dlmtr = '\\';
    if (pos == String::npos)
    {
        pos = filename.rfind('/');
        dlmtr = '/';
    }
    dirName = pos == string::npos ? "" : filename.substr(0, pos) + dlmtr;


    while( !file.eof() )
    {
        string str; getline( file, str );
        if( str.empty() ) break;
        trainFilenames.push_back(str);
    }
    file.close();
}


static bool createDetectorDescriptorMatcher( const string& detectorType, const string& descriptorType, const string& matcherType,
                                      Ptr<FeatureDetector>& featureDetector,
                                      Ptr<DescriptorExtractor>& descriptorExtractor,
                                      Ptr<DescriptorMatcher>& descriptorMatcher )
{
    cout << "< Creating feature detector, descriptor extractor and descriptor matcher ..." << endl;
    featureDetector = FeatureDetector::create( detectorType );
    descriptorExtractor = DescriptorExtractor::create( descriptorType );
    descriptorMatcher = DescriptorMatcher::create( matcherType );
    cout << ">" << endl;


    bool isCreated = !( featureDetector.empty() || descriptorExtractor.empty() || descriptorMatcher.empty() );
    if( !isCreated )
        cout << "Can not create feature detector or descriptor extractor or descriptor matcher of given types." << endl << ">" << endl;


    return isCreated;
}


static bool readImages( const string& queryImageName, const string& trainFilename,
                 Mat& queryImage, vector <Mat>& trainImages, vector<string>& trainImageNames )
{
    cout << "< Reading the images..." << endl;
    queryImage = imread( queryImageName, CV_LOAD_IMAGE_GRAYSCALE);
    if( queryImage.empty() )
    {
        cout << "Query image can not be read." << endl << ">" << endl;
        return false;
    }
    string trainDirName;
    readTrainFilenames( trainFilename, trainDirName, trainImageNames );
    if( trainImageNames.empty() )
    {
        cout << "Train image filenames can not be read." << endl << ">" << endl;
        return false;
    }
    int readImageCount = 0;
    for( size_t i = 0; i < trainImageNames.size(); i++ )
    {
        string filename = trainDirName + trainImageNames[i];
        Mat img = imread( filename, CV_LOAD_IMAGE_GRAYSCALE );
        if( img.empty() )
            cout << "Train image " << filename << " can not be read." << endl;
        else
            readImageCount++;
        trainImages.push_back( img );
    }
    if( !readImageCount )
    {
        cout << "All train images can not be read." << endl << ">" << endl;
        return false;
    }
    else
        cout << readImageCount << " train images were read." << endl;
    cout << ">" << endl;


    return true;
}


static void detectKeypoints( const Mat& queryImage, vector<KeyPoint>& queryKeypoints,
                      const vector<Mat>& trainImages, vector<vector<KeyPoint> >& trainKeypoints,
                      Ptr<FeatureDetector>& featureDetector )
{
    cout << endl << "< Extracting keypoints from images..." << endl;
    featureDetector->detect( queryImage, queryKeypoints );
    featureDetector->detect( trainImages, trainKeypoints );
    cout << ">" << endl;
}


static void computeDescriptors( const Mat& queryImage, vector<KeyPoint>& queryKeypoints, Mat& queryDescriptors,
                         const vector<Mat>& trainImages, vector<vector<KeyPoint> >& trainKeypoints, vector<Mat>& trainDescriptors,
                         Ptr<DescriptorExtractor>& descriptorExtractor )
{
    cout << "< Computing descriptors for keypoints..." << endl;
    descriptorExtractor->compute( queryImage, queryKeypoints, queryDescriptors );
    descriptorExtractor->compute( trainImages, trainKeypoints, trainDescriptors );


    int totalTrainDesc = 0;
    for( vector<Mat>::const_iterator tdIter = trainDescriptors.begin(); tdIter != trainDescriptors.end(); tdIter++ )
        totalTrainDesc += tdIter->rows;


    cout << "Query descriptors count: " << queryDescriptors.rows << "; Total train descriptors count: " << totalTrainDesc << endl;
    cout << ">" << endl;
}


static void matchDescriptors( const Mat& queryDescriptors, const vector<Mat>& trainDescriptors,
                       vector<DMatch>& matches, Ptr<DescriptorMatcher>& descriptorMatcher )
{
    cout << "< Set train descriptors collection in the matcher and match query descriptors to them..." << endl;
    TickMeter tm;


    tm.start();
    descriptorMatcher->add( trainDescriptors );
    descriptorMatcher->train();
    tm.stop();
    double buildTime = tm.getTimeMilli();


    tm.start();
    descriptorMatcher->match( queryDescriptors, matches );
    tm.stop();
    double matchTime = tm.getTimeMilli();


    CV_Assert( queryDescriptors.rows == (int)matches.size() || matches.empty() );


    cout << "Number of matches: " << matches.size() << endl;
    cout << "Build time: " << buildTime << " ms; Match time: " << matchTime << " ms" << endl;
    cout << ">" << endl;
}


static void saveResultImages( const Mat& queryImage, const vector<KeyPoint>& queryKeypoints,
                       const vector<Mat>& trainImages, const vector<vector<KeyPoint> >& trainKeypoints,
                       const vector<DMatch>& matches, const vector<string>& trainImagesNames, const string& resultDir )
{
    cout << "< Save results..." << endl;
    Mat drawImg;
    vector<char> mask;
    for( size_t i = 0; i < trainImages.size(); i++ )
    {
        if( !trainImages[i].empty() )
        {
            maskMatchesByTrainImgIdx( matches, (int)i, mask );
            drawMatches( queryImage, queryKeypoints, trainImages[i], trainKeypoints[i],
                         matches, drawImg, Scalar(255, 0, 0), Scalar(0, 255, 255), mask );
            string filename = resultDir + "/res_" + trainImagesNames[i];
            if( !imwrite( filename, drawImg ) )
                cout << "Image " << filename << " can not be saved (may be because directory " << resultDir << " does not exist)." << endl;
        }
    }
    cout << ">" << endl;
}


int main(int argc, char** argv)
{
    string detectorType = defaultDetectorType;
    string descriptorType = defaultDescriptorType;
    string matcherType = defaultMatcherType;
    string queryImageName = defaultQueryImageName;
    string fileWithTrainImages = defaultFileWithTrainImages;
    string dirToSaveResImages = defaultDirToSaveResImages;

    cv::initModule_nonfree();


    if( argc != 7 && argc != 1 )
    {
        printPrompt( argv[0] );
        return -1;
    }


    if( argc != 1 )
    {
        detectorType = argv[1]; descriptorType = argv[2]; matcherType = argv[3];
        queryImageName = argv[4]; fileWithTrainImages = argv[5];
        dirToSaveResImages = argv[6];
    }


    Ptr<FeatureDetector> featureDetector;
    Ptr<DescriptorExtractor> descriptorExtractor;
    Ptr<DescriptorMatcher> descriptorMatcher;
    if( !createDetectorDescriptorMatcher( detectorType, descriptorType, matcherType, featureDetector, descriptorExtractor, descriptorMatcher ) )
    {
        printPrompt( argv[0] );
        return -1;
    }


    Mat queryImage;
    vector<Mat> trainImages;
    vector<string> trainImagesNames;
    if( !readImages( queryImageName, fileWithTrainImages, queryImage, trainImages, trainImagesNames ) )
    {
        printPrompt( argv[0] );
        return -1;
    }


    vector<KeyPoint> queryKeypoints;
    vector<vector<KeyPoint> > trainKeypoints;
    detectKeypoints( queryImage, queryKeypoints, trainImages, trainKeypoints, featureDetector );


    Mat queryDescriptors;
    vector<Mat> trainDescriptors;
    computeDescriptors( queryImage, queryKeypoints, queryDescriptors,
                        trainImages, trainKeypoints, trainDescriptors,
                        descriptorExtractor );


    vector<DMatch> matches;
    matchDescriptors( queryDescriptors, trainDescriptors, matches, descriptorMatcher );


    saveResultImages( queryImage, queryKeypoints, trainImages, trainKeypoints,
                      matches, trainImagesNames, dirToSaveResImages );
    return 0;
}

一切正常。搞定。