Extracting selected text by bounding box from an image

Question

I am trying to fetch selected text by bounding box on an Image. like if only on word is selected by bounding box and I want to fetch that text and convert it into the text file. Please see my code and give some review so I can implement that functionality.

So far what I've done I've converted the PDF file to image with bounding box over the text.

import numpy as np
import csv
import io
from PIL import Image
import pytesseract
from wand.image import Image as wi
from pytesseract import Output
import cv2

pdf = wi(filename="samplecompany.pdf", resolution=100)
pdfImg = pdf.convert('jpg')
j = 1
for img in pdfImg.sequence:
    page = wi(image=img)
    page.save(filename=str(j)+".jpg")
    img1 = cv2.imread(str(j)+".jpg")

    d = pytesseract.image_to_data(img1, output_type=Output.DICT)
    n_boxes = len(d['level'])
    print(n_boxes)
    for i in range(n_boxes):
        (x, y, w, h) = (d['left'][i], d['top']
                        [i], d['width'][i], d['height'][i])
        print((x, y, w, h))
        cv2.rectangle(img1, (x, y), (x + w, y + h), (0, 255, 0), 2)

    cv2.imwrite(str(j)+".jpg", img1)

    cv2.waitKey(0)
    j += 1

this code is working fine I need to fetch desired text from images which I've created.using bounding box location

Answer 1

You can use this code to get custom text from a an image and change and modify accordingly and this is also save your text to an text file

import io
import cv2
import numpy as np
import pytesseract
from PIL import Image
from pytesseract import Output
from wand.image import Image as wi
import sys


pdf = wi(filename="Resume.pdf", resolution=100)
pdfImg = pdf.convert('jpg')
j = 1
imgBlobs = []
img1= []
for img in pdfImg.sequence:
    page = wi(image=img)
    page.save(filename=str(j)+".jpg")
    img1.append(cv2.imread(str(j)+".jpg"))
    j += 1

extracted_text = []

for img2 in img1:
    d = pytesseract.image_to_data(img2, output_type=Output.DICT)
    n_boxes = len(d['level'])
    print(n_boxes)
    extracted_text.append(d['text'][9])
    (x, y, w, h) = (d['left'][9], d['top'][9], d['width'][9], d['height'][9])
    cv2.rectangle(img2, (x, y), (x + w, y + h), (0, 255, 0), 2)


    cv2.imshow('img', img2)

    print(d)


with open('Prototype.txt', 'w') as filehandle:
        for listitem in extracted_text:
            filehandle.write('%s\n' % listitem)

Answer 2

let r denotes the bounding box (x, y, w, h) of your target. the croping can be done by

# Crop image
Croped_image = Image[int(r[1]):int(r[1]+r[3]), int(r[0]):int(r[0]+r[2])]

So you have the bounding box. Then try to detect the text from the segmented croped image. Whatever side is what you wanted