Is your feature request related to a problem?
I sometimes want to keep the colors in documents. Also, sometimes especially with very small text, the original image is readable, but the binary black-and-white image is not readable, and has very sharp ugly pixel edges. This is because the process removes all the greyscale colors at the edges.
Describe the Desired Solution
The quickest solution to implement for now could be a high-pass filter: create a 2nd, very strongly blurred version of the image. Use that to normalize the image, like described in https://superuser.com/a/1856394
Describe Alternatives You've Considered
There are also a lot of other algorithms with much better results.
One example (but without specification): https://sdk.docutain.com/blogartikel/scanner-app-filter
Some neural-networks based research:
https://github.com/xiaoyu258/DocProj
https://github.com/ZZZHANG-jx/GCDRNet
https://github.com/harrytea/UDoc-GAN
Compatibility with Design Philosophy
Is your feature request related to a problem?
I sometimes want to keep the colors in documents. Also, sometimes especially with very small text, the original image is readable, but the binary black-and-white image is not readable, and has very sharp ugly pixel edges. This is because the process removes all the greyscale colors at the edges.
Describe the Desired Solution
The quickest solution to implement for now could be a high-pass filter: create a 2nd, very strongly blurred version of the image. Use that to normalize the image, like described in https://superuser.com/a/1856394
Describe Alternatives You've Considered
There are also a lot of other algorithms with much better results.
One example (but without specification): https://sdk.docutain.com/blogartikel/scanner-app-filter
Some neural-networks based research:
https://github.com/xiaoyu258/DocProj
https://github.com/ZZZHANG-jx/GCDRNet
https://github.com/harrytea/UDoc-GAN
Compatibility with Design Philosophy