How to find the same ROI for the same text with different colors?

Question

I am trying to find ROIs on these two images:

I'm using this code for image #1:

image_1 = image1
corr1 = []
gray = cv2.cvtColor(image_1, cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray, (1,1), 1)
thresh = cv2.adaptiveThreshold(blur,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY_INV,11,10)

kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3,3))
dilate = cv2.dilate(thresh, kernel, iterations=3)

cnts = cv2.findContours(dilate, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
ROI_numbers1 = 0
ROI1 = []
for c in cnts:
    area = cv2.contourArea(c)
    if area > 5:
        x,y,w,h = cv2.boundingRect(c)
        cv2.rectangle(image_1, (x, y), (x + w, y + h), (0,255,0), 1)
        ROI1.append(image_1[y:y+h, x:x+w])
        corr1.append([y,y+h, x,x+w])
        ROI_numbers1 += 1

And this code for image #2:

image_2 = image2
corr2 = []
gray = cv2.cvtColor(image_2, cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray, (1,1), 1)
thresh = cv2.adaptiveThreshold(blur,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY_INV,11,10)

kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3,3))
dilate = cv2.dilate(thresh, kernel, iterations=3)

cnts = cv2.findContours(dilate, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
ROI_numbers2 = 0
ROI2 = []
for c in cnts:
    area = cv2.contourArea(c)
    if area > 5:
        x,y,w,h = cv2.boundingRect(c)
        cv2.rectangle(image_2, (x, y), (x + w, y + h), (0,255,0), 1)
        ROI2.append(image_2[y:y+h, x:x+w])
        corr2.append([y,y+h, x,x+w])
        ROI_numbers2 += 1

After using OpenCV for displaying the ROIs, I am getting this:

Why is the ROI area for the blue text in image #1 less than for the white text in image #2?

@HansHirse can you suggest some solution if possible. – anim esh Mar 29 '21 at 12:00 — anim esh, Mar 29 '21 at 12:00

HansHirse · Accepted Answer · 2021-04-01T10:35:58.417

When converting your image(s) to grayscale, you'll get different gray values for the white and blue texts. Thus, cv2.GaussianBlur will give different results, and following cv2.adaptiveThreshold, too. In the end, the found contours are different, following the ROIs.

Don't convert to grayscale here! In your original, three-channel image, mask anything that's not the background, which is a solid gray (53, 53, 53). That mask replaces your thresh. From there, you can then use your existing implementation.

Here's a minimal example to check, whether the resulting bounding rectangles (ROIs) are the same:

import cv2
import numpy as np


def cnts_from_image(image):
    thresh = (~np.all(image == (53, 53, 53), axis=2)).astype(np.uint8) * 255
    kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3, 3))
    dilate = cv2.dilate(thresh, kernel, iterations=3)
    cnts = cv2.findContours(dilate, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
    cnts = cnts[0] if len(cnts) == 2 else cnts[1]
    return cnts


rects_white = [cv2.boundingRect(c) for c in cnts_from_image(cv2.imread('white_text.png'))]
rects_blue = [cv2.boundingRect(c) for c in cnts_from_image(cv2.imread('blue_text.png'))]

print('All rectangles identical:', np.all([rw == rb for rw, rb in zip(rects_white, rects_blue)]))
# All rectangles identical: True

----------------------------------------
System information
----------------------------------------
Platform:      Windows-10-10.0.16299-SP0
Python:        3.9.1
NumPy:         1.20.2
OpenCV:        4.5.1
----------------------------------------

Its working all good for gray background but not for any other. Should I manually enter different values in (53, 53, 53) to check for different background colors? — anim esh, Apr 03 '21 at 04:03
@animesh Exactly, you need to adapt this RGB/BGR value according to the background colour. — HansHirse, Apr 03 '21 at 14:01

How to find the same ROI for the same text with different colors?

1 Answers1