False positive finder pattern detection inside QR code

[lessthanoptimal]

As discussed lessthanoptimal/PyBoof#23 there are false positive finder pattern inside the data portion of the QR code. This causes the graph to be formed incorrectly and it fails to decode the QR. These are perfect QR and it shouldn't have any issues.

Tasks:

• Add a new regression test where it decodes 5000 or so QR
• Fix the problem and see if it reduces the number of failures.

[simon-staal]

Just a quick update on this, I've tested all values in the range $r \in [0, 10^6]$ exhaustively, and the following payloads are problematic:

[2474, 4916, 6079, 7243, 7531, 7763, 8138, 8698, 9049, 9158, 9372, 9884, 10110, 10427, 11207, 11303, 11674, 12389, 14244, 14515, 16312, 16456, 18454, 19705, 20415, 20436, 21145, 21749, 21914, 24172, 25165, 25660, 25997, 26322, 26405, 27887, 28255, 29747, 30368, 30726, 33405, 33820, 34006, 35119, 35318, 36409, 37855, 38863, 39885, 41031, 41097, 41688, 42746, 42841, 43425, 50792, 51712, 51764, 51795, 53224, 54567, 55074, 55446, 55522, 55577, 55643, 55881, 57067, 58467, 58561, 59293, 60299, 60857, 60983, 62159, 62861, 64115, 64376, 65192, 65330, 65482, 65699, 65910, 66539, 67010, 67614, 67880, 68250, 68365, 68620, 68752, 69920, 69997, 70120, 70292, 70570, 70708, 70715, 71009, 72280, 72420, 73730, 75275, 75295, 75897, 76185, 77126, 79383, 79737, 79929, 81720, 82106, 83475, 83630, 84087, 84088, 84409, 84848, 86459, 87442, 91170, 91265, 91582, 92247, 92950, 93117, 94101, 94182, 94914, 97164, 99174, 99224]

Could be helpful for testing.

[lessthanoptimal]

thanks for posting these.