How to detect edge and crop an image in Python

Question

I'm new to Image Processing in Python and I'm trying to solve a common problem. I have an image having a signature of a person. I want to find the edges and crop it to fit the signature in the image.

Input Image

Expected Output

I tried Canny Edge Detection and cropping the image using a list of existing solutions (articles & answers) using PIL, CV2, but none seem to work. I'm looking for a working solution.

Some solutions I tried:

1. https://www.quora.com/How-can-I-detect-an-object-from-static-image-and-crop-it-from-the-image-using-openCV

2. Crop Image from all sides after edge detection

3. How to crop biggest rectangle out of an image

and many more... None worked although seems very simple. I encountered either errors or not expected output using any of the existing solutions.

Answer 1

What you need is thresholding. In OpenCV you can accomplish this using cv2.threshold().

I took a shot at it. My approach was the following:

1. Convert to grayscale
2. Threshold the image to only get the signature and nothing else
3. Find where those pixels are that show up in the thresholded image
4. Crop around that region in the original grayscale
5. Create a new thresholded image from the crop that isn't as strict for display

Here was my attempt, I think it worked pretty well.

import cv2
import numpy as np

# load image
img = cv2.imread('image.jpg') 
rsz_img = cv2.resize(img, None, fx=0.25, fy=0.25) # resize since image is huge
gray = cv2.cvtColor(rsz_img, cv2.COLOR_BGR2GRAY) # convert to grayscale

# threshold to get just the signature
retval, thresh_gray = cv2.threshold(gray, thresh=100, maxval=255, type=cv2.THRESH_BINARY)

# find where the signature is and make a cropped region
points = np.argwhere(thresh_gray==0) # find where the black pixels are
points = np.fliplr(points) # store them in x,y coordinates instead of row,col indices
x, y, w, h = cv2.boundingRect(points) # create a rectangle around those points
x, y, w, h = x-10, y-10, w+20, h+20 # make the box a little bigger
crop = gray[y:y+h, x:x+w] # create a cropped region of the gray image

# get the thresholded crop
retval, thresh_crop = cv2.threshold(crop, thresh=200, maxval=255, type=cv2.THRESH_BINARY)

# display
cv2.imshow("Cropped and thresholded image", thresh_crop) 
cv2.waitKey(0)

And here's the result: