Barcorama - barcode reader software [Updated 11/Oct/09]

[pizzaguy]

I get the same error as nsa666 on my tilt (default rom)

[srowen]

Originally Posted by hypest EAN13 generally seems to work. Please, could you try this image? I have tested the prog on about 10 actual products. Try to keep the phone about 6-10cm away from the target (not closer than 5, anyway). Thank you for the info on the behavior of the load dialog. I will try and use a different facility to browse every folder. hypest

In the interest of speed it doesn't by default try real hard to find a barcode. It will only find 1D codes that are "upright" -- long axis is basically left to right. Some rotation is OK. So the image you posted won't get recognized. It also won't work if flipped upside down.

Both of those are quite possible, just not desirable by default. On a slow phone, where we're trying 6 1D formats and QR Codes, we don't want to spend too much time. We were thinking of adding a "try harder" hint/mode that would tell it to try a longer search.

The other note is that it will look around the vertical middle of the image for a barcode but not search too far outside that range. So you should try to center the barcode in the capture. Again that could be changed or expanded in a "try harder" mode.

You can toss images against the decoder library here:
http://zxing.org/w/decode.jspx

The original image won't decode but it will if you flip it 90 degrees clockwise.

[hypest]

to srowen:
I have actually slightly modified the recognition code to search both horizontally and vertically, though only "left to right" in both cases. So the photo with the "green cd case" I've uploaded does get recognized. Have you tried it and it failed?

hypest

[hypest]

Originally Posted by nsa666 P.S.2: just tested the verbose version. It seems to be something wrong with directory name. it says \Storagecard\Barcorama\Barcorama_verbose.exe. I think it should say only \Storagecard\Barcorama .

nsa666:
This observation was certainly good! I've made some adjustments and I've uploaded (yet another build) in #1 post - the "nonewdir" one. Perhaps you want to try that also...

hypest

[nsa666]

Great, new version works! Now you can try to speed it up( for example settings for codetypes that should be recognized schould greatly improve the performance). If it was my programm i would extract the recognition in another thread(than the graphics) and make a progress bar or something, because it seems for some seconds that it hangs.

[hypest]

Originally Posted by nsa666 Great, new version works! Now you can try to speed it up( for example settings for codetypes that should be recognized schould greatly improve the performance). If it was my programm i would extract the recognition in another thread(than the graphics) and make a progress bar or something, because it seems for some seconds that it hangs.

At last! It was rather frustrating that the prog worked only on my device!

Your recommendation about the separate thread is actually already in the version I'm working on

hypest

[pizzaguy]

Sweet! After many, many, tries realizing it was on the wrong resolution or i was holding it the wrong way, it works!


...has anyone tried it with 2D barcodes?

[srowen]

Originally Posted by hypest to srowen: I have actually slightly modified the recognition code to search both horizontally and vertically, though only "left to right" in both cases. So the photo with the "green cd case" I've uploaded does get recognized. Have you tried it and it failed? hypest

Ah no I was trying with the most recent version of our (Java) code. Makes sense that it works if you've modified it to check other orientations.

[srowen]

Originally Posted by nsa666 Great, new version works! Now you can try to speed it up( for example settings for codetypes that should be recognized schould greatly improve the performance). If it was my programm i would extract the recognition in another thread(than the graphics) and make a progress bar or something, because it seems for some seconds that it hangs.

Not sure exactly how much the code was modified, but as of version 0.3, you should be able to easily select a Reader implementation that only looks for the format you want.

EAN13Reader will only look for EAN13, and should be blazingly fast by itself. MultiFormatOneDReader will look for all six supported 1D formats pretty efficiently. If you use the full MultiFormatReader then it's also looking for QR Codes (later, Data Matrix too) which you don't seem to want. But at least it tries 1D formats first, so, you're only incurring that overhead when a 1D format is not recognized.

[hypest]

Please, find in the #1 post a newer version, implementing new (micro)features and incorporating the fixes from the "remoting debugging" sessions I had here with fellow posters! This is a more user-friendly version as it bypasses the second "ok" to close the camera application, and uses backgound threads for the main jobs.

to srowen:
I have already spotted the decoder-selection functionality, and also have noticed the oftenly quick response of the EAN13 decoder. I hope I will quickly make progress in the basic application functionality so I can dive into the zxing code. I'm particularily interested in experimenting with the image preprocessing.

have fun,
hypest