Hough algorithm - thinning lines out

Hough

asked 2018-12-27 15:32:47 -0600

Filippo9669
31 ●1 ●5

Hello. FunctionHoughLines()returns a vector filled with these of lines which get enough votes during the voting process. Now, let's suppose: I have three lines very close to each other, so the differences in angles are very small, about 1 or 2 degree, so:

Theta = {20, 24, 25}

How can I manage to choose the one that is the most certain to be a line that I'm looking for (e.x the longest one)? To be specific: with the highest amount of votes out of these three above? What shall I compare?

edit retag flag offensive close merge delete

add a comment

1 answer

Sort by » oldest newest most voted

answered 2019-10-07 11:35:16 -0600

nicolasabril
59 ●1 ●3

In c++, if you pass a Vec3f as the output vector, the third "dimension" is the amount of votes that line got. You can classify the lines using a clustering algorithm like DBSCAN and then choose only the line with the most votes inside each cluster.
Another way of deciding the angle is taking the average of the rhos and thetas of each cluster. This works better if you have a lot of lines in each cluster.