0

I want to calculate BER (bit error rate) of two different QR codes. Can't find the variable in the decoder provide by ZXING which stored the bit stream form the decoder prior to the error correction decoding procedure.

Rishabh1906
  • 1
  • 2
  • Is there a way to access the raw data before correction? If so, you could simply xor raw image with corrected image and count the 1 bits after the xor. Note that not all of the QR image is protected by error correction code, and the format info is 15 to 30 bits long with it's own bit oriented error correction code, and depending on the QR Code it may be simpler to do a table lookup for the format info instead of doing error correction. – rcgldr Apr 03 '18 at 10:00
  • Yeah, I was reading the patent and found that if we could unmask the QR code corresponding to the error correction level and mask being applied we may get the raw data bits encoded in the QR. I am working on it. I think doing so I can look for the differences in the module/bits of two QR codes excluding the standard format module. That would be a good measure to calculate BER. Thanks for the suggestion. – Rishabh1906 Apr 04 '18 at 10:54
  • "calculate BER" - do you mean calculate the maximum BER that a QR code can correct? You can calculate the maximum byte error rate based on the correction code. Given this, you can then calculate a bit error rate, but the calculation is a bit tricky, as a pair of bad bits could end up in the same byte. For example, say a code can correct 6 bytes out of 32 bytes. It turns out that 7 bits in error will end up in 6 out of 32 bytes more than 50% of the time. The maximum BER is complicated by probabilities of distribution of error bits. – rcgldr Apr 04 '18 at 16:15
  • I need to find the efficiency of a certain algorithm which develops new QR code with some alteration. For this, I am interested in knowing the overall Bit Error Rate being introduced due to modification. As the algorithm changes the codeword so comparing the codeword bitwise is a good measure to know how much the new QR code has changed. – Rishabh1906 Apr 05 '18 at 06:23
  • All though there are two methods for this sort of comparison i.e either I directly compare the two codes without digging into the abstract codeword or I compare the actual codeword bitwise. The latter seems to be more promising as for different mask and error correction level the codewords changes a lot and the error introduced may vary abruptly. – Rishabh1906 Apr 05 '18 at 06:27
  • The issue here is the error correction code is byte based. It would be simpler to compare error correction as a percentage of bytes, This is already approximated in the error correction level, (L M Q H), you could calculated the exact percentage for a specific error correction code. This can then be translated into a byte error rate, which in turn can be translated into a bit error rate. – rcgldr Apr 06 '18 at 23:46

0 Answers0