I'm trying to iterate through end points of lines, build segments between them and finally dewarp the distortion. I should imagine, that all those "mini"-segments are small images (somthing about 70x300, but it variets.)
This what I done:
//allPoints are created in another class, but the output
//for std::vector> allPoints is as follows:
//(for the first:)[139, 113][201, 310][223, 339][297, 437][323, 472][381, 465][408, 413][484, 291][505, 151]
//(for the second:)[139, 210][201, 692][223, 664][297, 550][323, 523][381, 544][408, 580][484, 699][505, 254]
for (int i = 0; i < allPoints.size() - 1; i++) {
cv::Vec3b zero(0, 0, 0);
cv::Mat_ img_dst(allPoints[i].second.y - allPoints[i].first.y,
allPoints[i+1].first.x - allPoints[i].first.x);
std::cout << "\nSize of img_dst: " << img_dst.size() << "\n";
const cv::Point2f src_pt[] = {
cv::Point2f(allPoints[i].first),
cv::Point2f(allPoints[i].second),
cv::Point2f(allPoints[i + 1].second),
cv::Point2f(allPoints[i + 1].first) };
const cv::Point2f dst_pt[] = {
cv::Point2f(allPoints[i].first),
cv::Point2f(allPoints[i].second),
cv::Point2f(allPoints[i + 1].second.x, allPoints[i].second.y),
cv::Point2f(allPoints[i + 1].first.x, allPoints[i].first.y) };
const cv::Mat homography_matrix = cv::getPerspectiveTransform(src_pt, dst_pt);
cv::warpPerspective(matCopy, img_dst, homography_matrix, img_dst.size());
cv::imshow("Result", img_dst);
std::pair, std::string> wallNr = WallParser::wallParser(argument);
int label = wallNr.first[i];
std::cout << label << " ";
Main::saveAll(img_dst, "", label);
}
So actually I'm giving the right size and also the right points, which should be dewrap. But I get instead of (created it quicklin ps - in PS it results a bigger image) [![()][1]][1] this one [![enter image description here][2]][2]. The image looks like this and all those end points are the end points of the yellow lines:
[![enter image description here][3]][3]
How can I get that `wrapPrespective` to dewrap right?
[1]: https://i.stack.imgur.com/QpSkk.jpg
[2]: https://i.stack.imgur.com/rG2Qg.png
[3]: https://i.stack.imgur.com/Do77n.jpg
↧