[libdmtx-open_discussion] Using libdmtx on constrained ARM device (i.e. Nokia N900)

[Simon P. <S.G...@ba...>]

Hello chaps,

I'm the author of mbarcode, a barcode scanning app for the Nokia N900. I use
ZBar for 1D and QR code decoding and your libdmtx for Datamatrix decoding.

Datamatrix is comparatively slow on the N900 (compared with QR decoding for
example), which is doubtless partly because QR has nice big recognition
symbols which make alignment of the barcode far easier, but I also wanted to
check and see what options people would recommend using to improve
performance?

I'm actually in the process of testing Onion's gst-dmtx in combination with
Stefan Kost's gst-zbar, in the hope that I can move all of the heavy lifting
out to the Gstreamer thread(s) and then perhaps write the UI in Python (to
make it easier for people to add plugins and the like).

The data I'm feeding to the decoder is 320 x 240 pixels Y800 (though 640 x
480 is also available if that would improve things) and currently I simply
set the decoder timeout. I see that there are lots of options (e.g.
http://git.tal.org/gitweb.cgi?p=gst-dmtx.git;a=blob;f=gst-dmtx-test-pipe.sh;
h=72951bdd9326fcee47475da61a5808e06aca4a51;hb=HEAD), what would people
recommend and most importantly, is there a list describing these anywhere
(e.g. what does "skip" do and does "scale=2" shrink the image by a factor of
2?) 

Thanks,


Simon