gati nueta
However, zxing fails to detect any QR code on some of the images (geht_nicht_qr.png), while it seems to have no problems on the others (geht_qr.png).
Any ideas what's wrong with geht_nicht_qr.png?
Thank you.
Sean Owen
gati nueta
> These are actually some of the hardest images because lots of phantom barcodes and finder patterns are easy to find in this type of image. It doesn't find the third finder pattern in the first image, not quite. I can't get it to scan off screen either as it's a little to dense and slightly distorted. You may find very slightly different images work fine.
Thank you for looking into this.
The problem we have is that we are trying to detect QR codes from images from a document scanner, so we only have a single "bullet" recognizing the QR code. We won't be able to remove the phantoms, as we're not generating the codes ourselves.
About the slight distortions, we problably won't be able to eliminate them as they are caused by the scanner hardware - but we're unsure if this means that recognition with zxing is impossible.
In some cases the three finder patterns are correctly found, but the pixels in the finder patterns lead to a miscalculation of the module size and the dimension is incorrectly calculated (Detector.computeDimension throws NotFoundException or returns wrong result).
It seems also in other cases the module size is not correctly estimated leading to a rejection of correct finder patterns - I can debug into this but I'm not sure if it will lead anywhere...
Do you think there is anything we can do to increase our chances of recognizing the QR codes?
gjs
I could scan both images ok from computer screen using zxing app running on a Google Android Pixel XL phone device. Nicht image returned UPC code 16033522 but no idea if that code is correct or not...