How to find “regions” in a contour OpenCV?

Question

Say we have the following contour information from OpenCV contours:

What I mean by a "region" is a subset of the contour with low directional variation.

So for example these, could be regions in the provided example:

One way to detect these could be, doing a local neighborhoud comparison of the dot products of the tangent at each point. (ie see how much the tangent changes locally).

I was wondering however if there is a better way to do this, using OpenCV directly rather than doing vector operations myself.

Answer 1

-When your region boundaries are always near-vertical or near-horizontal, consider preprocessing the image using a filter (erode, dilate), to isolate vertices and horices, then merge results, to find an alternating color on region boundaries.

-When your directions go anywhere, it's more complicated ! One option would be to retrieve coordinates from your pixels with the help of Hough lines see

https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_imgproc/py_houghlines/py_houghlines.html