Ok, I read the documentation and how Randomized Hough Transform (let's call it RHT) works, and I have this idea. Perhaps this will be wrong.
In RHT, not all binary points are taken into account, correctly. That is why it is faster. Thus, it thresholdmeans in the battery how many voices you had to consider (rho, theta) as a line, it's simple.
, , minLineLength , , . , 5 , , lineLength , 5 . , 3 , - , lineLength 20 .
, Length, .