Detecting and Localising The Region In Two Images That Differentiate With Text (Plain/Curve/Rotate)

user_13680864 0
Viewed 80

I've two identical images, only differences are one of them contain some text. I would like to make a bounding box or polygonal box around that text. To do that, I tried to compare these two images and measure the difference area and next make a bounding box around the different area or simply the text region.

Original Images

Text + Original Images


before = cv2.imread('before.jpg')
after = cv2.imread('after.jpg')

# Convert images to grayscale
before_gray = cv2.cvtColor(before, cv2.COLOR_BGR2GRAY)
after_gray = cv2.cvtColor(after, cv2.COLOR_BGR2GRAY)

# Compute SSIM between two images
(score, diff) = compare_ssim(before_gray, after_gray, full=True)
print("Image similarity", score)

# The diff image contains the actual image differences between the two images
# and is represented as a floating point data type in the range [0,1] 
# so we must convert the array to 8-bit unsigned integers in the range
# [0,255] before we can use it with OpenCV
diff = (diff * 255).astype("uint8")

# Threshold the difference image, followed by finding contours to
# obtain the regions of the two input images that differ
thresh = cv2.threshold(diff, 0, 255, cv2.THRESH_BINARY_INV | cv2.THRESH_OTSU)[1]
contours = cv2.findContours(thresh.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
contours = contours[0] if len(contours) == 2 else contours[1]

or c in contours:
    area = cv2.contourArea(c)
    if area > 40:
        x,y,w,h = cv2.boundingRect(c)
        cv2.rectangle(before, (x, y), (x + w, y + h), (36,255,12), 2)
        cv2.rectangle(after, (x, y), (x + w, y + h), (36,255,12), 2)

The issue is, it fails to capture with the bounding box of the rotate text and curve shape text. Please, see the image below:

Comparison Results

Wanted

I would like to achieve a proper bounding box or if possible polygonal box around the text whether it is rotated or curve. An expected demo output is given below:

Expected Result

opencv image-processing python
2 Answers

You need to calculate the point-sequence for each text location.

for c in contours:
    area = cv2.contourArea(c)
    if area > 40:
        x, y, w, h = cv2.boundingRect(c)
        rectangle = cv2.minAreaRect(c)
        box = np.int0(cv2.boxPoints(rectangle))
        cv2.drawContours(after, [box], 0, (0, 191, 255), 2)

Output:

enter image description here

user_54042260

Its an easy task, Use the minAreaRect Here is the documentation
For example,

rect = cv2.minAreaRect(cnt)
box = cv2.BoxPoints(rect).astype(np.int)
cv2.drawContours(im,[box],0,(0,0,255),2)


user_132095890
Related