rqrr not able to group capstone of image

[sergentbne]

rqrr is not able to detect this QRcode

here is the function that uses rqrr

fn decode_img_with_data(img_data: image::ImageBuffer<Luma<u8>, Vec<u8>>) -> Vec<u8> {
    // Prepare for detection

    let mut img = rqrr::PreparedImage::prepare(img_data);
    // Search for grids, without decoding
    let grids = img.detect_grids();

    let mut data = Vec::new();
    // Decode the grid
    let _ = grids[0].decode_to(&mut data).unwrap();
    assert!(data.len() != 0);
    return data;
}

and the results

thread 'main' panicked at src/decode_from_mp4.rs:162:18:
index out of bounds: the len is 0 but the index is 0
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

idk if other information is required, so it would be so nice of you to help me!

also! the qrcodes are always of the same size, so is there a way to just "declare" a grid instead of detecting it?

[mohnishsalian0]

The qr is corrupt I believe. I tried using this with other scanners and didn't work

[sergentbne]

that seems the most likely option, I will try to fix it and return if the problem persist!

Thank you for your time!

[sergentbne]

Zbar is able to read the QRcode in full, without any corruption. The problem may stem from the fact that this QRcode is raw binary data. Although I am not sure if the problem is from me or the library, it would be nice to know how to manually define a qrcode for decoding

[mohnishsalian0]

Actually you are right. Usually scanners convert binary to string, therefore the UTF conversion was failing since its raw binary. However, in this case the scanner completely fails to detect the qr.

After investigation it looks like it manages to detect the capstones but fails to group them.
I believe this line could be responsible:
const VIABILITY_THRESHOLD: f64 = 0.25;

If I'm correct, in the previous version this threshold was 2.5. Not sure why this was changed.

@WanzenBug Any thoughts on this?

BTW, I'm not an official contributor, just a fellow enthusiast. I can raise a PR if you wish

[sergentbne]

Changing back the VIABILITY_THRESHOLD to 2.5 as it used to be fixes the problem. Perhaps it is a viable option for the future?

[WanzenBug]

Changing back the VIABILITY_THRESHOLD to 2.5 as it used to be fixes the problem. Perhaps it is a viable option for the future?

It does allow it to detect the grid, but it still can't be decoded. I also tested another v40 grid from wikipedia: https://commons.wikimedia.org/wiki/File:Qr-code-ver-40.svg

That also does not seem to work :/

[mohnishsalian0]

It does allow it to detect the grid, but it still can't be decoded.

That's because it is not a valid UTF sequence. So string conversion doesn't work. One solution for this would be to provide another interface:
decode_to_bytes(&self) -> DeQRResult<(MetaData, Vec<u8>)>

https://commons.wikimedia.org/wiki/File:Qr-code-ver-40.svg

I tried the png version (440px x 440px) of this, and it worked.