Experiment I: pattern reliability.
Effective pattern reliability means that patterns with lower reliability are more likely to fail the scan in reality than patterns with higher reliability.
To verify this hypothesis, we uniformly divided the range of reliability into four intervals and classified patterns into four groups. For each group, we generated 60 synthetic QR codes using the patterns from the group, and each QR code was scanned by three users using four barcode readers. Figure 7(a) shows the success rates of barcode readers
operating on each group. It indicates not only an expected positive correlation between pattern reliability and response of barcode readers, but also the relative performance among barcode readers.
In the next experiment, we shows that by adjusting the value of parameter λ, we could effectively control the readability of generated halftone QR codes.