Original file looks like that:
https://i.stack.imgur.com/qPj7I.jpg
I need to detect columns that contain numeric data for further processing.
I used the following method to get my columns:
# Detect vertical lines
kernel = np.ones((35,1), np.uint8) # note this is a horizontal kernel
d_im = cv2.dilate(image_channels, kernel, iterations=1)
verticle_lines_img = cv2.erode(d_im, kernel, iterations=5)
cv2.imwrite("images/outputc3.jpg",verticle_lines_img)
Thats the result:
https://i.stack.imgur.com/S95BK.jpg
Now based on this image I would like to split the original one into subfiles where each file is a jpg or png image containing only one table. Based on the fact that the image is partly skewed one column image does not have to contain pieces of other table data.