如何从图像中提取垂直和水平线

Question

Task:

I have a set of images where every image looks like this:

I'd like to extract all horizontal and all vertical lines from this image.

Desired results:

Current approach:

import cv2


image = cv2.imread('img.png')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray, (3, 3), 0)
thresh = cv2.threshold(blur, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]

horizontal_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (20, 1))
horizontal = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, horizontal_kernel, iterations=1)

Here's the result:

Problem:

It's clear that the current kernel is too narrow to bypass this thick vertical line on the left. This line is 41-pixel thick, so the kernel (42, 1) works fine, but I lose true horizontal lines that a shorter than 41 pixel:

Are there any flawless techniques for solving this problem?

Answer 1

The idea is to bring all those lines to the same "Size" by Skeletonization before applying a morphological filter & use a smaller filter Size

image = cv2.imread('Your_ImagePath' , cv2.IMREAD_COLOR)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]

thresh = skeletonize(thresh)
cv2.imshow("Skelton",thresh)
cv2.waitKey()

horizontal_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (10, 1))
horizontal = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, horizontal_kernel, 
iterations=1)

cv2.imshow("Horizontal Lines" , horizontal)
v2.waitKey()

OUTPUT IMAGE:

NOTE: Skeletonization code from skeletonize.py

Answer 2

I would be thinking about doing a "Morphological Opening" with a tall, thin structuring element (ie a vertical line 15 pixels tall and 1 pixel wide) to find the vertical bars and with a horizontal line 15 pixels long and 1 pixel high to find the horizontal bars.

You can do it just the same with OpenCV . but I am just doing it here with ImageMagick in the Terminal because I am quicker at that: Here's a command for the vertical work - try varying the 15 to get different results:

magick shapes.png -morphology open rectangle:1x15 result.png

Here's an animation of how the result changes as you vary the 15 (the length of the line):

And here is how it looks when you make the structuring element a horizontal line:

magick shapes.png -morphology open rectangle:15x1 result.png

If you are new to morphology, there is an excellent description by Anthony Thyssen here . Note that it may be explained in terms of ImageMagick but the principles are equally applicable in Python OpenCV - see here .