Crop simple bounding box with openCV

Question

I have some images with a black background and some text in the corner:

I'm trying to do a rectangular crop to make it look like:

The text on the sides as well as the window dimensions are of varying sizes. My code isn't cropping correctly, what am I doing wrong?

I've tried removing the text in the bottom right corner first and cropping, that doesn't work either.

def crop_cont(img):
     gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
     _,thresh = cv2.threshold(gray,15,255,cv2.THRESH_BINARY)

     _, contours, _= cv2.findContours(thresh, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
     cnt = contours[0]

     x,y,w,h = cv2.boundingRect(cnt)

     crop = img[y:y+h,x:x+w]

return crop

Answer 1

Your code is in general ok. The issue is that you are using contours[0]. You have to find the right contour (there are more than one). In this particular example, the right contour is the biggest one. Just iterate all found contours and find the one with the biggest area.

Answer 2

An approach is to determine the contour with the largest area then crop around this contour. We can do this by converting the image to grayscale, performing canny edge detection, then some morphological operations to smooth the image. From here we find contours and sort using contour area. To crop the ROI, we simply determine the bounding rectangle coordinates and use Numpy slicing. Here's the result

import cv2

image = cv2.imread('1.png')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
canny = cv2.Canny(gray, 120, 255, 1)
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5,5))
close = cv2.morphologyEx(canny, cv2.MORPH_CLOSE, kernel)

cnts = cv2.findContours(close, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
cnts = sorted(cnts, key=cv2.contourArea)[-1:]
for c in cnts:
    x,y,w,h = cv2.boundingRect(c)
    ROI = image[y:y+h,x:x+w]

cv2.imshow('ROI', ROI)
cv2.waitKey()