Extract individual field from table image to excel with OCR

Question

I have scanned images which have tables as shown in this image:

I am trying to extract each box separately and perform OCR but when I try to detect horizont

Answer 1

nanthancy's answer is also accurate, I used the following script for getting each box and sorting it by columns and rows.

Note: Most of this code is from a medium blog by Kanan Vyas here: https://medium.com/coinmonks/a-box-detection-algorithm-for-any-image-containing-boxes-756c15d7ed26

#most of this code is take from blog by Kanan Vyas here: 
#https://medium.com/coinmonks/a-box-detection-algorithm-for-any-image-containing-boxes-756c15d7ed26


import cv2
import numpy as np

img = cv2.imread('images/scan2.jpg',0)

#fn to show np images with cv2 and close on any key press
def imshow(img, label='default'):
    cv2.imshow(label, img)
    cv2.waitKey(0)
    cv2.destroyAllWindows()

# Thresholding the image
(thresh, img_bin) = cv2.threshold(img, 250, 255,cv2.THRESH_BINARY|cv2.THRESH_OTSU)
#inverting the image
img_bin = 255-img_bin 

# Defining a kernel length
kernel_length = np.array(img).shape[1]//80

# A verticle kernel of (1 X kernel_length), which will detect all the verticle lines from the image.
verticle_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (1, kernel_length))# A horizontal kernel of (kernel_length X 1), which will help to detect all the horizontal line from the image.
hori_kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (kernel_length, 1))# A kernel of (3 X 3) ones.
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3, 3))

# Morphological operation to detect vertical lines from an image
img_temp1 = cv2.erode(img_bin, verticle_kernel, iterations=3)
verticle_lines_img = cv2.dilate(img_temp1, verticle_kernel, iterations=3)
#cv2.imwrite("verticle_lines.jpg",verticle_lines_img)

# Morphological operation to detect horizontal lines from an image
img_temp2 = cv2.erode(img_bin, hori_kernel, iterations=3)
horizontal_lines_img = cv2.dilate(img_temp2, hori_kernel, iterations=3)
#cv2.imwrite("horizontal_lines.jpg",horizontal_lines_img)

# Weighting parameters, this will decide the quantity of an image to be added to make a new image.
alpha = 0.5

beta = 1.0 - alpha# This function helps to add two image with specific weight parameter to get a third image as summation of two image.

img_final_bin = cv2.addWeighted(verticle_lines_img, alpha, horizontal_lines_img, beta, 0.0)

img_final_bin = cv2.erode(~img_final_bin, kernel, iterations=2)

(thresh, img_final_bin) = cv2.threshold(img_final_bin, 128,255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)

cv2.imwrite("img_final_bin.jpg",img_final_bin)


# Find contours for image, which will detect all the boxes
contours, hierarchy = cv2.findContours(img_final_bin, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)

""" this section saves each extracted box as a seperate image.
idx = 0
for c in contours:
    # Returns the location and width,height for every contour
    x, y, w, h = cv2.boundingRect(c)
    #only selecting boxes within certain width height range
    if (w > 10 and h > 15 and h < 50):
        idx += 1
        new_img = img[y:y+h, x:x+w]
        #cv2.imwrite("kanan/1/"+ "{}-{}-{}-{}".format(x, y, w, h) + '.jpg', new_img)
"""

#get set of all y-coordinates to sort boxes row wise
def getsety(boxes):
    ally = []
    for b in boxes:
        ally.append(b[1])
    ally = set(ally)
    ally = sorted(ally)
    return ally

#sort boxes by y in certain range, because if image is tilted than same row boxes 
#could have different Ys but within certain range
def sort_boxes(boxes, y, row_column):
    l = []
    for b in boxes:
        if (b[2] > 10 and b[3] > 15 and b[3] < 50):
            if b[1] >= y - 7 and b[1] <= y + 7:
                l.append(b)

    if l in row_column:
        return row_column
    else:
        row_column.append(l)
        return row_column

#sort each row using X of each box to sort it column wise
def sortrows(rc):
    new_rc = []
    for row in rc:
        r_new = sorted(row, key = lambda cell: cell[0])
        new_rc.append(r_new)
    return new_rc

row_column = []
for i in getsety(boundingBoxes):
    row_column = sort_boxes(boundingBoxes, i, row_column)
row_column = [i for i in row_column if i != []]

#final np array with sorted boxes from top left to bottom right
row_column = sortrows(row_column)

I made this in Jupyter notebook and copy-pasted here, if any errors come up, let me know.

Thank you everyone for answers

Answer 2

This is function, which uses tesseract-ocr for layout detection. You can try with different RIL levels and PSM. For more details have a look here: https://github.com/sirfz/tesserocr

import os
import platform
from typing import List, Tuple

from tesserocr import PyTessBaseAPI, iterate_level, RIL

system = platform.system()
if system == 'Linux':
    tessdata_folder_default = ''
elif system == 'Windows':
    tessdata_folder_default = r'C:\Program Files (x86)\Tesseract-OCR\tessdata'
else:
    raise NotImplementedError

# this tesseract specific env variable takes precedence for tessdata folder location selection
# especially important for windows, as we don't know if we're running 32 or 64bit tesseract
tessdata_folder = os.getenv('TESSDATA_PREFIX', tessdata_folder_default)


def get_layout_boxes(input_image,  # PIL image object
                     level: RIL,
                     include_text: bool,
                     include_boxes: bool,
                     language: str,
                     psm: int,
                     tessdata_path='') -> List[Tuple]:
    """
    Get image components coordinates. It will return also text if include_text is True.
    :param input_image: input PIL image
    :param level: page iterator level, please see "RIL" enum
    :param include_text: if True return boxes texts
    :param include_boxes: if True return boxes coordinates
    :param language: language for OCR
    :param psm: page segmentation mode, by default it is PSM.AUTO which is 3
    :param tessdata_path: the path to the tessdata folder
    :return: list of tuples: [((x1, y1, x2, y2), text)), ...]
    """
    assert any((include_text, include_boxes)), (
        'Both include_text and include_boxes can not be False.')

    if not tessdata_path:
        tessdata_path = tessdata_folder

    try:
        with PyTessBaseAPI(path=tessdata_path, lang=language) as api:
            api.SetImage(input_image)

            api.SetPageSegMode(psm)
            api.Recognize()
            page_iterator = api.GetIterator()
            data = []
            for pi in iterate_level(page_iterator, level):
                bounding_box = pi.BoundingBox(level)
                if bounding_box is not None:
                    text = pi.GetUTF8Text(level) if include_text else None
                    box = bounding_box if include_boxes else None
                    data.append((box, text))
            return data
    except RuntimeError:
        print('Please specify correct path to tessdata.')

Answer 3

You're on the right track. Here's a continuation of your approach with slight modifications. The idea is:

1. Obtain binary image. Load image, convert to grayscale, and Otsu's threshold.

2. Remove all character text contours. We create a rectangular kernel and perform opening to only keep the horizontal/vertical lines. This will effectively make the text into tiny noise so we find contours and filter using contour area to remove them.

3. Repair horizontal/vertical lines and extract each ROI. We morph close to fix and broken lines and smooth the table. From here we sort the box field contours using imutils.sort_contours() with the top-to-bottom parameter. Next we find contours and filter using contour area then extract each ROI.

---

Here's a visualization of each box field and the extracted ROI

Code

import cv2
import numpy as np
from imutils import contours

# Load image, grayscale, Otsu's threshold
image = cv2.imread('1.jpg')
original = image.copy()
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]

# Remove text characters with morph open and contour filtering
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (3,3))
opening = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, kernel, iterations=1)
cnts = cv2.findContours(opening, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
for c in cnts:
    area = cv2.contourArea(c)
    if area < 500:
        cv2.drawContours(opening, [c], -1, (0,0,0), -1)

# Repair table lines, sort contours, and extract ROI
close = 255 - cv2.morphologyEx(opening, cv2.MORPH_CLOSE, kernel, iterations=1)
cnts = cv2.findContours(close, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
cnts = cnts[0] if len(cnts) == 2 else cnts[1]
(cnts, _) = contours.sort_contours(cnts, method="top-to-bottom")
for c in cnts:
    area = cv2.contourArea(c)
    if area < 25000:
        x,y,w,h = cv2.boundingRect(c)
        cv2.rectangle(image, (x, y), (x + w, y + h), (36,255,12), -1)
        ROI = original[y:y+h, x:x+w]

        # Visualization
        cv2.imshow('image', image)
        cv2.imshow('ROI', ROI)
        cv2.waitKey(20)

cv2.imshow('opening', opening)
cv2.imshow('close', close)
cv2.imshow('image', image)
cv2.waitKey()