Measuring angle of a curve wire with OpenCV

Question

I have a robotic device that bends wire, and I try to measure whether the bending was successful by looking at the desired and true angle measurements.

I'm very new to image processing so please excuse my crude terminology.

What I have in mind is to detect the wire and draw a line for each segment of the wire, after that it becomes just a little trigonometry problem. But I can not find a way to draw and get coordinates of those lines.

Answer 1

I would try these steps:

1. Skeletonize the image.
2. Get the relevant points from the skeleton-path.
3. Calculate the angle.

EDIT:

import cv2
import numpy as np
from skimage import morphology, graph
from skan import Skeleton

def angle(v1, v2):
    rad = np.arctan2(v2[0], v2[1]) - np.arctan2(v1[0], v1[1])
    return np.abs((np.rad2deg(rad)))

# read an threshold image
img = cv2.imread('img.png')
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
inverted = 255 - gray
ret, thresh = cv2.threshold(inverted, 50, 255, cv2.THRESH_BINARY)

# skeletonize image and split skeleton to paths, for now assume there is only one
skeleton = morphology.skeletonize(thresh, method='lee')
g = Skeleton(skeleton)
path = np.array(g.path_coordinates(0).astype(int))

# draw path for demonstration purpose
for p in path:
    img[p[0], p[1]] = [0, 0, 255]

# get reference points by percental position on path
l = len(path)
a1 = path[int(0.02 * l)]
a2 = path[int(0.10 * l)]
b1 = path[int(0.98 * l)]
b2 = path[int(0.90 * l)]

# results
print(f"Degrees: {angle(a1 - a2, b1 - b2):0.2f}")
cv2.line(img, (a1[1], a1[0]), (a2[1], a2[0]), [0, 255, 0], 2)
cv2.line(img, (b1[1], b1[0]), (b2[1], b2[0]), [0, 255, 0], 2)
cv2.imwrite("out.png", img)

Output:

Degrees: 32.32