2d Ridge Peak Detection // Extracting Line Segments from a Density Plot Image using OpenCV

Question

I have density plots for line locations. When visualizing the density plot image, multiple lines can be observed, each composed of one or more segments. I'm looking for a solution that would allow me to extract these lines and represent them as a list of points (e.g., [start_point, p2, p3, end_point]).

Is there an existing Python OpenCV (cv2) function or an open-source implementation that can help me achieve this? I would greatly appreciate any guidance or suggestions.

The density plot showing the line locations. (i.e.: what I have)

The line segments drawn with a line width of 1 pixel. (i.e.: what I want)

One approach I thought of is a brute-force method, where I would loop over all x's and then loop over all y's to detect the "humps" of each line segment and join them with nearby humps from the previous x. Then, I would merge segments that are near linear. I would repeat the process, looping over all y's and then x's, to detect vertical lines. However, I'm looking for a more efficient solution if possible.

Thank you in advance for your help!

Answer 1

fmw42 comment is the perfect solution.

Threshold. Then get contours. Then draw each one separately as white filled on black. Then thin using skimage skeletonize (scikit-image.org/docs/stable/auto_examples/edges/…) on the drawn contour. Then get the points along the skeleton for each contour. –