fazlurnu/text-extraction-table-image

Text-Extraction-Table-Image

This project aims to extract text from a table image into python objects. Below is a result of the detection:

Prerequisites/Dependencies

• OpenCV => 2.4.8
• Numpy
• PyTesseract

Idea Behind The Code

I've publisehed the documentation on my website. Please read it to understand the idea behind the code.

For Refinement

After your algorithm can detect the text successfully, now you can save it into Python object such as Dictionary or List. Some regions name (in the “Kabupaten/Kota” are failed to be detected precisely, since it is not included in Tesseract training data. However, it shouldn’t be a problem as the regions’ indexes can be detected precisely. Also, this text extraction might fail to detect the text in other fonts, depending on the font used. In case of misinterpretation, such as “5” is detected as “8”, you can do an image processing such as eroding and dilating.

My code is far from perfect, if you find some error or chances of refinement, write me a comment!

text-extraction-table-image's Issues

List index out of range

Hi there thanks for sharing this wonderful algo, but i'm gtting this error..

File "c:\Program Files\Sublime Text 3\Text-Extraction-Table-Image-master\scripts\ROI_selection.py", line 113, in get_ROI
x1 = vertical[left_line_index][2] + offset
IndexError: list index out of range

images/source.png is jpg file

images/source.png has wrong file extension - it is jpg file and not png.
And main.py is expecting to find source7.png, so IMO this should be changed to:

diff --git a/scripts/main.py b/scripts/main.py
index de29972..d6dad3d 100644
--- a/scripts/main.py
+++ b/scripts/main.py
@@ -10,7 +10,7 @@ from ROI_selection import detect_lines, get_ROI
 import cv2 as cv

 def main(display = False, print_text = False, write = False):
-    filename = '../images/source7.png'
+    filename = '../images/source.jpg'

     src = cv.imread(cv.samples.findFile(filename))

Ignore