I’m looking for an algorithm that detects the cell bounds in a borderless table.
This is the result I’m looking to achieve:
The algorithm should return a list of rectagle borders (in a particular order) so I can use them to segment the image into cells and run an OCR on each cell.
Any help would be very appreciated.
opencv has a text detection thingy. “EAST” or something. that gives you boxes.
now that you have the boxes, you need to associate them to each other as a “graph” (general case). left/right/up/down. now you can work with that graph (navigate around in it, find the top left cell for example), which should ideally have grid shape. if not, there’s something funky going on.