How to draw a line (line with rectangular corners) with OpenCV and python

user_16983791 0
Viewed 100

How can I draw a line with rectangular corners with OpenCV line, I don't want to use OpenCV's rectangle for some reason. I have posted my code below and output as well, In the output line is curved at the endpoint and start points, I require it as a rectangle.

I searched a lot for that, but it seems no one had that problem before.

import cv2
thickness = 24
img1 = cv2.imread("AMIR4.jpg")
cv2.line(img1, (190,55), (85,55), (255,0,0), (thickness))
cv2.imshow("image", img1)
cv2.waitKey(0)

Sample:

enter image description here

opencv image-processing shapes python
2 Answers

According to OpenCV docs, line() function will draw thick lines with rounding endings.

That being said, you cannot directly get over this issue. You could draw a line several times with thickness=1:

cv2.line(img1, (190, 55), (85, 55), (255, 0, 0), 1)
cv2.line(img1, (190, 56), (85, 56), (255, 0, 0), 1)
cv2.line(img1, (190, 57), (85, 57), (255, 0, 0), 1)
...

And although you're reluctant, I would suggest simply drawing a filled rectangle instead.

user_146648610

I can see how the cv2.rectangle() method wouldn't work for you if you want the lines to be able to have any rotation.

So here is a function I defined that works just like the cv2.line() method, but will result in rectangular ends rather than circular ends:

def draw_line(img, pt1, pt2, color, thickness):
    x1, y1, x2, y2 = *pt1, *pt2
    theta = np.pi - np.arctan2(y1 - y2, x1 - x2)
    dx = int(np.sin(theta) * thickness / 2)
    dy = int(np.cos(theta) * thickness / 2)
    pts = [
        [x1 + dx, y1 + dy],
        [x1 - dx, y1 - dy],
        [x2 - dx, y2 - dy],
        [x2 + dx, y2 + dy]
    ]
    cv2.fillPoly(img, [np.array(pts)], color)

Testing alongside the cv2.line() method:

import cv2
import numpy as np

def draw_line(img, pt1, pt2, color, thickness):
    x1, y1, x2, y2 = *pt1, *pt2
    theta = np.pi - np.arctan2(y1 - y2, x1 - x2)
    dx = int(np.sin(theta) * thickness / 2)
    dy = int(np.cos(theta) * thickness / 2)
    pts = [
        [x1 + dx, y1 + dy],
        [x1 - dx, y1 - dy],
        [x2 - dx, y2 - dy],
        [x2 + dx, y2 + dy]
    ]
    cv2.fillPoly(img, [np.array(pts)], color)

img = np.full((600, 600, 3), 255, np.uint8)

pt1 = np.random.randint(0, 600, 2)
pt2 = np.random.randint(0, 600, 2)

draw_line(img, pt1, pt2, (0, 0, 0), 15)
cv2.line(img, pt1, pt2, (255, 255, 255), 1)

cv2.imshow("Image", img)
cv2.waitKey(0)

Output (will vary because of the np.random.randint()):

enter image description here

user_135524700
Related