OpenCV: Add text to image and get the bounding box of each letter

Question

I read the documentation of the OpenCV function putText(). I could not find there any parameter or returned value that tells where each printed letter is located.

Is there some OpenCV API that gives this data while adding text to images, or maybe in some different library?

You want to write text to image and then find contours of that text? — zindarod, Aug 31 '17 at 13:25
No. I want something more accurate than `findCountours()` - the text could be handwritten-like, such that contours won't do the job. I simply want to know the bounding box of each character that was printed on the image — SomethingSomething, Aug 31 '17 at 13:26
But are you doing the printing or the text is already printed on the image? — zindarod, Aug 31 '17 at 13:27
Maybe there is an option to imitate what `putText()` does, then, while controlling what's printed, saving all the boxes to somewhere? — SomethingSomething, Aug 31 '17 at 13:27
I am doing the printing myself, and I want to know the bounding box of each character — SomethingSomething, Aug 31 '17 at 13:27

score 1 · Answer 1 · answered Aug 31 '17 at 13:33

1

You can calculate the width and height of the text you're putting on image by:

cv::Size textSize = cv::getTextSize(text, fontFace, fontScale, thickness, &baseline);

Check for example here.

answered Aug 31 '17 at 13:33

zindarod

6,328
3
30
58

It calculates the width and height of a whole string, HOWEVER, given this information, I would be able to print them char by char and always update the offset with the last char's width. Thank you! It may be very helpful! – SomethingSomething Aug 31 '17 at 13:36
@SomethingSomething Yes that's the idea. Glad I could help. – zindarod Aug 31 '17 at 13:37
For hand-written text, it may disconnect the letters, but it may not be that harmful for my needs – SomethingSomething Aug 31 '17 at 13:38
@SomethingSomething If you post an image with text of your choice, it would give us a better idea. – zindarod Aug 31 '17 at 13:39

score 1 · Answer 2 · answered Sep 01 '17 at 14:07

1

Take a look at my answer, the code may be interesting for your purpose: Detect space between text (OpenCV, Python)

It is used to recognize handwriting text and do ROI on given image.

answered Sep 01 '17 at 14:07

lucians

2,239
5
36
64

OpenCV: Add text to image and get the bounding box of each letter

2 Answers2