FieldImageProcessor.cpp 7.09 KB
Newer Older
Thiago Santini's avatar
Thiago Santini committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177
#include "FieldImageProcessor.h"
#include <opencv2/highgui.hpp>

using namespace std;
using namespace cv;
using namespace aruco;

static int gFieldDataId = qRegisterMetaType<FieldData>("FieldData");

FieldImageProcessor::FieldImageProcessor(QString id, QObject *parent)
    : id(id),
      sROI(QPointF(0,0)),
      eROI(QPointF(1,1)),
      forceSanitize(false),
      QObject(parent)
{
    settings = new QSettings(gCfgDir + "/" + id + " ImageProcessor", QSettings::IniFormat);
    updateConfig();

    dict = getPredefinedDictionary(DICT_4X4_250);
    detectorParameters = new DetectorParameters();
    detectorParameters->markerBorderBits = 2;
    detectorParameters->minMarkerPerimeterRate = 0.10; // TODO: determine a good value for these based on the fov and maximum detection distance
    //printMarkers(); // TODO: parametrize me
}

void FieldImageProcessor::updateConfig()
{
    QMutexLocker locker(&cfgMutex);
    cfg.load(settings);
    forceSanitize = true;
}

FieldImageProcessor::~FieldImageProcessor()
{
    if(settings)
        settings->deleteLater();
}

void FieldImageProcessor::process(Timestamp timestamp, const Mat &frame)
{
    Mat prevInput = data.input;

    // TODO: parametrize frame drop due to lack of processing power
    if (gTimer.elapsed() - timestamp > 50)
        return;

    QMutexLocker locker(&cfgMutex);

    data.timestamp = timestamp;

    // Always force the creation of a new matrix for the input since the old one might still be alive from further down the pipeline
    if (cfg.inputSize.width > 0 && cfg.inputSize.height > 0) {
        data.input = Mat(cfg.inputSize, frame.type() );
        resize(frame, data.input, cfg.inputSize);
    }
    else
        data.input = frame.clone();

    if (cfg.flip != CV_FLIP_NONE)
        flip(data.input, data.input, cfg.flip);

    sanitizeCameraParameters( Size(data.input.cols, data.input.rows) );

    data.undistorted = cfg.undistort;
    if (data.undistorted)
        remap(data.input, data.input, map1, map2, CV_INTER_AREA);

    data.width = data.input.cols;
    data.height = data.input.rows;

    // Marker detection and pose estimation
    vector<int> ids;
    vector<vector<Point2f> > corners;
    if (cfg.processingDownscalingFactor > 1) {
        Mat downscaled;
        resize(data.input, downscaled, Size(),
               1/cfg.processingDownscalingFactor,
               1/cfg.processingDownscalingFactor,
               INTER_AREA);
        if (cfg.markerDetectionMethod == "aruco") {
            detectMarkers(downscaled, dict, corners, ids, detectorParameters);
            for (unsigned int i=0; i<ids.size(); i++)
                for (unsigned int j=0; j<corners[i].size(); j++)
                    corners[i][j] = cfg.processingDownscalingFactor*corners[i][j];
        }
    } else
        if (cfg.markerDetectionMethod == "aruco")
            detectMarkers(data.input, dict, corners, ids, detectorParameters);

    // Filling the marker data
    data.collectionMarker = Marker();
    data.markers.clear();

    // Note that the following is based on the COLLECTION MARKER size.
    if (ids.size() > 0) {
        if (data.undistorted) {
            Mat emptyCameraMatrix = (Mat_<double>(3,3) <<
                           data.width,          0, data.width/2,
                           0         , data.width, data.height/2,
                           0         ,0          , 1);
            Mat emptyDistCoeffs = (Mat_<double>(1,4) << 0, 0, 0, 0);
            estimatePoseSingleMarkers(corners, cfg.collectionMarkerSizeMeters, emptyCameraMatrix, emptyDistCoeffs, rvecs, tvecs);
        } else
            estimatePoseSingleMarkers(corners, cfg.collectionMarkerSizeMeters, cameraMatrix, distCoeffs, rvecs, tvecs);
    }

    for (unsigned int i=0; i<ids.size(); i++) {
        Marker marker(corners[i], ids[i]);

        marker.center = estimateMarkerCenter(marker.corners);
        marker.center.z = tvecs.at<double>(i,2);
        marker.tv = ( Mat_<float>(1,3) << tvecs.at<double>(i,0), tvecs.at<double>(i,1), tvecs.at<double>(i,2) );
        marker.rv = ( Mat_<float>(1,3) << rvecs.at<double>(i,0), rvecs.at<double>(i,1), rvecs.at<double>(i,2) );

        data.markers.push_back( marker );

        // use closest calibration marker -- to try and avoid detecting the one viewed in the field camera when testing :-)
        if (marker.id == cfg.collectionMarkerId) {
            if (data.collectionMarker.id == -1)
                data.collectionMarker = data.markers.back();
            else if (data.collectionMarker.center.z > marker.center.z)
                data.collectionMarker = data.markers.back();
        }
    }

    data.validGazeEstimate = false;
    data.processingTimestamp = gTimer.elapsed();

    Q_ASSERT_X(prevInput.data != data.input.data, "Field Image Processing", "Previous and current input image matches!");
    emit newData(data);
}

void FieldImageProcessor::newROI(QPointF sROI, QPointF eROI)
{
    QMutexLocker locker(&cfgMutex);
    if (sROI.isNull() || eROI.isNull()) {
        this->sROI = QPointF(0,0);
        this->eROI = QPointF(1,1);
    } else {
        this->sROI = sROI;
        this->eROI = eROI;
    }
}

void FieldImageProcessor::sanitizeCameraParameters(Size size)
{
    if (!forceSanitize) {
        if ( (size == imageSize) && !map1.empty() && !map2.empty() )
            return;
    } else
        forceSanitize = false;

    FileStorage fs( QString(gCfgDir + "/" + id + "Calibration.xml").toStdString(), FileStorage::READ);
    fs["cameraMatrix"] >> cameraMatrix;
    fs["distCoeffs"] >> distCoeffs;
    fs["imageSize"] >> imageSize;
    bool haveCameraParameters = false;
    if (!cameraMatrix.empty() && !distCoeffs.empty())
        haveCameraParameters = true;

    double currentAspectRatio = size.width / (float) size.height;
    double originalAspectRatio = imageSize.width / (float) imageSize.height;

    if (haveCameraParameters && (currentAspectRatio == originalAspectRatio) ) {
        qInfo() << "Found intrinsic parameters.";
        double rx = imageSize.width  / (double) size.width;
        double ry = imageSize.height / (double) size.height;
        cameraMatrix.at<double>(0,2) = cameraMatrix.at<double>(0,2) / rx;
        cameraMatrix.at<double>(1,2) = cameraMatrix.at<double>(1,2) / ry;
        cameraMatrix.at<double>(0,0) = cameraMatrix.at<double>(0,0) / rx;
        cameraMatrix.at<double>(1,1) = cameraMatrix.at<double>(1,1) / rx;
        imageSize = size;
    } else {
        qInfo() << "No valid intrinsic parameters available. Using dummy values";
        // Dummy estimation
        cameraMatrix = (Mat_<double>(3,3) <<
Thiago Santini's avatar
Thiago Santini committed
178 179
		                size.width, 0, 0.5*size.width,
		                0, size.width, 0.5*size.height,
Thiago Santini's avatar
Thiago Santini committed
180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
                        0, 0 ,1 );
        distCoeffs = (Mat_<double>(1,4) << 0, 0, 0, 0);
        imageSize = size;
    }

    initUndistortRectifyMap(
                            cameraMatrix,
                            distCoeffs,
                            Mat(),
                            getOptimalNewCameraMatrix(cameraMatrix, distCoeffs, imageSize, 0, imageSize),
                            imageSize,
                            CV_32FC1,
                            map1,
                            map2
                            );
}