# Hough transform logic

Dear all,

I got couple of questions regarding Hough Transform for line detection:

1) I have read on many places that it is preferable (and in most cases done by) to use polar coordinates rather than cartesian. What is the reason behind that?

2) Voting: As I got it, we already have edge points in the image (by applying some kind of edge detector before)

Then we go to one of pixels which is known to be part of edge, get its "rho" value (distance to origin) and check for all theta values if there is any match with other edge pixels along imaginary (rho, theta) line.

-> If there is match - we get one vote, then we do this for all edge pixels and as soon as any (rho, theta) combination passes threshold for number of votes it is considered as a line.

Respectively, in Hough space above procedure will mean that if there is more than certain amount of line intersections at one point then this point represents line.

Question: Is my understanding of procedure correct or I am missing something important here?

Sorry if my questions sound a bit dumb, but I would like to get a good sense of Hough transform logic and so far I am not sure if I am getting it right.

Bests, Jev