vipul-sharma20/document-scanner

An OpenCV based document scanner

document-scanner's Issues

Will it work for all images

I gave a cheque as input but your program is not working for that, Can you please help on this...

cv2 has not member imread and resize

I just tried to debug the scanner.py file and it tells me in line 8, cv2 has no 'imread' member:
Has anyone had the same problem yet? Am I missing a module or package that needs to be installed?
I already installed opencv-python succesfully...
Thanks

blog post

I noticed someone in the HN thread posted about how a blog post from 2014 had the same code as found in this repository.

We should give that pyimagesearch blog post credit in the README

Modify perspective rectification to match paper aspect ratio

Currently, when the code performs the perspective rectification/correction, the code maps the original image to an [800 x 800] image (800 is hardcoded).

As a result, the resulting image looks "squashed" in the x direction.

To improve the output, one could map the image to an image size whose aspect ratio matches the paper size aspect ratio.
For instance, let's assume that the form will come in the A4 paper size:
http://www.papersizes.org/a-paper-sizes.htm
The A4 ratio of height-to-width is 11.7 / 8.3 ~= 1.41.

So, to maintain an A4 aspect ratio when doing the perspective correction, you can map the original image to an image of size [800 x 1128](800 * 1.41 = 1128).

Since your test image is landscape, the dimensions should be:
Height: 800 pixels
Width: 1128 pixels
It'd be neat to also autodetect paper orientation, ie portrait vs landscape.

[Blog] "OpenCV's methods may not perform accurately for very large dimensions"

In your blog post:
http://vipulsharma20.blogspot.in/2016/01/document-scanner-using-python-opencv.html

You write that:
"The original image is resized and scaled down as OpenCV's methods may not perform accurately for very large dimensions. (Above image is the scaled down/resized version)"

I think you should consider rewording this sentence. Instead of "accuracy", do you mean performance/efficiency?

For instance, doing contour detection on a 18000 x 18000 image is likely to be extremely slow. Instead, one can downscale the image to a more manageable size -- say, 1800 x 1800 -- and still get comparable results.

Your original phrasing seems to suggest that certain operations will have bad results on very large images, but will somehow get better simply by downsizing the image. This seems odd -- and wrong -- to me.

Otherwise, I liked the writeup!

[Sorry, this issue doesn't deal with your code at all, but I thought I'd put it here to keep things localized. Thanks!]

contours error

Hello,

I get the following error with valid arguments and (to the best of my knowledge) satisfied dependencies. Any ideas?

Traceback (most recent call last):
File "/path/to/scanner.py", line 28, in
(contours, _) = cv2.findContours(edged, cv2.RETR_LIST, cv2.CHAIN_APPROX_NONE)
ValueError: too many values to unpack

Thank you!
zelon88