This is kind of a software/hardware question. I only know a little bit about what Lidar is capable of, but I keep expecting to see it integrated into iOS document scanners as a way to automatically correct for pages that aren’t perfectly flat, like open books, etc. Is the technology not really up for that? Are there iOS document scanners that already include that?
I’m not an expert on LiDAR but, my understanding is that LiDAR is a distance sensor that shoots out little beams of light and calculates the time it takes to be reflected back. So one could utilize LiDAR to detect curled pages and such but that’s completely separate from being able to do anything about it.
Not LiDAR bu this is a way to deal with all sorts of paper oddities
LiDAR on the iPhone is meant to improve the detection of spatial entities for improved Augemented Reality apps. So it could detect a curved page but not really do anything with that knowledge. My understanding is that it would not be able to detect any of the text on that page so that has to be done some other way.
Now you might look at the text unfolding algorithms references above. They are open source. They can take the images (in this case produced via CT scans) and reconfigure them so the text can be read. Then you’d have to apply a text recognition algorithm to the resulting images and hope for the best.
You also might look at this software
Or you could just buy one of these
Or the cheaper options like this
Yeah, I was imagining some software that could combine the two pieces of information – the 2-dimensional photo and the 3-dimensional depth map – and sort of uncurl the page. Perhaps it’s not viable.
1 Like
And why isn’t it? I would think it was.
Book scanners commonly have two cameras. In part this is to orient themselves close to perpendicular to the pages of an open book. Commonly books do not easily lie open completely flat. By having two cameras, one for each visible page, the cameras can be angled differently and each perpendicular to the page being filmed without the physical book being completely flat.
The other benefit of two cameras is that they can offer binocular vision. In this way, the shape of the curve at the spine of the book can be established by analysis of the two images. Then algorithms can correct for the resulting distortion in the images.
So book OCR readers can, in effect, “uncurl” the page without using LIDAR.
Your idea of LIDAR is intriguing. It certainly seems like such a thing would be possible. It might not be economically feasible.
3 Likes