如何在OPENCV C ++中将元素存储到向量的向量

Question

I have the vector "lines2" that I get from HoughLinesP transformation. And I have the distance transformation of the binary image. Lines2 only has the starting and ending points of the lines, so with LineIterator I go through all the lines and get the coordinates in between those points. I then use these coordinates to get the values from the distance image. Finally I would like to store all the values in "lines3", than when i++, store the next etc. Here is the code:

distanceTransform(img_bin, img_dist, CV_DIST_C, 3);
imshow("Distance Transform Image", img_dist);

//Go through the merged lines with LineIterator to get all the points between the endings
vector<vector<int>> lines3(lines2.size());
int val;

for (size_t i = 0; i < lines2.size(); i++) {

    LineIterator it(img_dist, Point(lines2[i][0], lines2[i][1]), Point(lines2[i][2], lines2[i][3]), 8);
    vector<Point> points(it.count);

    for (int j = 0; j < it.count; j++, ++it)
    {
        points[j] = it.pos();
        //cout << "points[j]: " << points[j] << endl;
        //check the value of pos() at the distance transformated image and store it in lines3
        val = img_dist.at<float>(points[j].y, points[j].x);
        //lines3(i,j)[0].push_back(val);
        //lines3[i][j].push_back(val);
        //lines3[i][j] = val;
        //lines3(i,j) = val;
        //lines3[j].push_back(val);
        lines3[i][j].push_back(val); //THIS LINE doeasn't even compile
    }
}

Question: how can I do that? The push_back is not working here.

EDIT: added some initial size to 'lines3'. But still don't know how can I fill the vector by [i][j] with the lines and values of lines.

lines3[j].push_back(val); //THIS LINE compiles but fails runtime with out of bounds error (of course)

Answer 1

I post this for the future, so maybe others won't stuck like me with this. As Dan Mašek mentioned in the comments I needed to replace

lines3[i][j].push_back(val);

with

lines3[i].push_back(val);

And now it works fine! Thanks!