I successfully detected the outer frame from the original floor plan, and the result is represented as a collection of points.
I’m trying to convert these points into line segments. I’ve already tried using direction-based splitting, DBSCAN clustering, and RANSAC line fitting, but none of them worked well. I suspect the failure is due to the points being too jagged or too densely packed, making it hard to form clean lines.
Do you recommend refining or simplifying the points first? Or would it be better to switch to using OpenCV-based methods?
Unfortunately, I don’t have an image to share, but since the floor plans vary, some parts are straight lines while others include curves. What approach would you recommend in this case?
I would really appreciate any ideas or suggestions from those with more experience. Thank you in advance!