void MarkerDetector::findContours(cv::Mat& thresholdImg,
ContoursVector& contours,
int minContourPointsAllowed)
{
ContoursVector allContours;
cv::findContours(thresholdImg, allContours, CV_RETR_LIST, CV_CHAIN_APPROX_NONE);
contours.clear();
for (size_t i=0; i<allContours.size(); i++)
{
int contourSize = allContours[i].size();
if (contourSize > minContourPointsAllowed)
{
contours.push_back(allContours[i]);
}
}
#ifdef SHOW_DEBUG_IMAGES
{
cv::Mat contoursImage(thresholdImg.size(), CV_8UC1);
contoursImage = cv::Scalar(0);
cv::drawContours(contoursImage, contours, -1, cv::Scalar(255), 2, CV_AA);
cv::imshow("Contours", contoursImage);
cv::waitKey();
}
#endif
}
bool MultiContourObjectDetector::findBaseShape(cv::Mat& baseImage)
{
vector<vector<vector<Point>>> approxContours = findApproxContours(baseImage, true, true);
if (approxContours.size() == 0)
{
cerr << "ERROR: No valid contours found in base image" << endl;
return false;
}
_baseShape = approxContours[0];
_aspectedContours = _baseShape.size();
_minContourPoints = numeric_limits<int>::max();
for (int i = 0; i < _baseShape.size(); i++)
{
vector<Point> tempKeypoints;
Utility::findCentroidsKeypoints(_baseShape[i], tempKeypoints, Utility::CentroidDetectionMode::THREE_LOOP);
_baseKeypoints.push_back(tempKeypoints);
if (_baseShape[i].size() < _minContourPoints)
_minContourPoints = _baseShape[i].size();
}
#ifdef DEBUG_MODE
Mat contoursImage(baseImage.size(), CV_8UC1);
contoursImage = cv::Scalar(0);
drawContours(contoursImage, _baseShape, -1, cv::Scalar(255), 1, CV_AA);
imshow("TO FIND CONTOUR", contoursImage);
#endif
return true;
}
// The workflow of the marker detection routine is the following:
// 1. convert the input image to grayscale;
// 2. perform a binary threshold operation;
// 3. detect contours;
// 4. search for possible markers;
// 5. detect and decode markers;
// 6. estimate marker 3D pose.
void MarkerDetector::processFrame(const cv::Mat& frame)
{
// ...
std::vector<Marker> markers;
//cv::Mat bgraMat( frame.rows, frame.cols, CV_8UC4 );
//cv::cvtColor( frame, bgraMat, CV_BGR2BGRA );
// Convert the image to grayscale.
// See: http://docs.opencv.org/modules/imgproc/doc/miscellaneous_transformations.html#cvtcolor
// The conversion to grayscale is necessary because markers usually contain only black and white blocks.
// So it is much easier to operate with them using grayscale images.
//cv::cvtColor( bgraMat, grayscaleImage, CV_BGRA2GRAY );
cv::cvtColor(frame, grayscaleImage, CV_BGR2GRAY);
// Make it binary.
// See: http://docs.opencv.org/modules/imgproc/doc/miscellaneous_transformations.html#adaptivethreshold
// The binarization operation will transform each pixel of our image to black (0 intensity) or white (255 intensity).
// This step is required to find contours.
cv::threshold(grayscaleImage, thresholdImg, 127.0, 255.0, cv::THRESH_BINARY_INV);
#ifdef _DEBUG
cv::imshow("THR", thresholdImg);
#endif
// Detect contours.
MarkerDetector::findContours(thresholdImg, contours, grayscaleImage.cols / 5);
#ifdef _DEBUG
cv::Mat contoursImage(thresholdImg.size(), CV_8UC1);
contoursImage = cv::Scalar(0);
// See: http://docs.opencv.org/modules/imgproc/doc/structural_analysis_and_shape_descriptors.html#drawcontours
cv::drawContours(contoursImage, contours, -1, cv::Scalar(255), 2, CV_AA);
cv::imshow("CNT", contoursImage);
#endif
// Find closed contours that can be approximated with 4 points.
findMarkerCandidates(contours, markers);
// Find if there are markers.
detectMarkers(grayscaleImage, markers);
// Calcualte their poses.
estimatePosition(markers);
// Sort by id.
// See: http://www.cplusplus.com/reference/algorithm/sort/
std::sort(markers.begin(), markers.end());
// ...
transformations.clear();
ids.clear();
for (size_t i = 0; i < markers.size(); i++) {
transformations.push_back(markers[i].transformation);
ids.push_back(markers[i].id);
}
}
// Each marker usually looks like a square figure with black and white areas inside it.
// So the best way to locate a marker is to find closed contours and approximate them with polygons of 4 vertices.
// This function will detect contours on the input binary image.
// The return value is a list of polygons where each polygon represents a single contour.
// This function skips contours that have their perimeter in pixel value set to be less than a specific input threshold.
// This because we are not interested in small contours.
void MarkerDetector::findContours(const cv::Mat& thresholdImg,
std::vector< std::vector<cv::Point> >& contours,
int minContourPointsAllowed)
{
std::vector< std::vector<cv::Point> > allContours;
// See: http://docs.opencv.org/modules/imgproc/doc/structural_analysis_and_shape_descriptors.html#findcontours
cv::findContours(thresholdImg, allContours, CV_RETR_LIST, CV_CHAIN_APPROX_NONE);
contours.clear();
for (size_t i = 0; i < allContours.size(); i++) {
int contourSize = allContours[i].size();
if (contourSize > minContourPointsAllowed) { contours.push_back(allContours[i]); }
}
#ifdef _DEBUG
cv::Mat contoursImage(thresholdImg.size(), CV_8UC1);
contoursImage = cv::Scalar(0);
cv::drawContours(contoursImage, contours, -1, cv::Scalar(255), 2, CV_AA);
cv::imshow("Contours Image", contoursImage);
#endif
}
