仅从图像 opencv 或 OCR 中提取字符

Question

From a group of text like below

I want to MAKE a BOUNDING BOX on INDIVIDUAL CHARACTER . However, I am unable to do so.

I've tried to use Easy OCR with following settings but it only recognizes individual words:

reader = eo.Reader(['en'],gpu=True)
result = reader.readtext(imgOriginal,y_ths=0.0000000001,x_ths=0.0000000001,paragraph=False)

I tried to set psm/oem in tesserocr/pytesserocr but still I wasn't able to get the individual character. Please Help.

Answer 1

Have a look at GetComponentImage example from tesserocr and adapt it:

from PIL import Image, ImageOps
from tesserocr import PyTessBaseAPI, RIL

image = ImageOps.grayscale(Image.open('test.png'))).convert('L')
with PyTessBaseAPI(path=tessdata_path, psm=tesserocr.PSM.SPARSE_TEXT) as api:
    api.SetImage(image)
    api.Recognize()
    boxes = api.GetComponentImages(RIL.SYMBOL, True)
    print('Found {} symbol image components.'.format(len(boxes)))
    for i, (im, box, _, _) in enumerate(boxes):
        print("Box[{0}]: x={x}, y={y}, w={w}, h={h}".format(i, **box))
        # display(im)

If boxes are not accurate try to use oem=tesserocr.OEM.TESSERACT_ONLY with correct trainneddata.