Watermark Detector Sensitivity via Quality-Based Weighted Decoding

The patent applies local quality by differentiating between individual repetitions based on their specific noise characteristics. Instead of treating all repetitions uniformly as in majority voting, the method evaluates each repetition's quality metric and applies weighted summation where each repetition contributes differently to the final decoded value based on its local quality assessment. This resolves the contradiction by maintaining computational simplicity while improving correction capacity through localized quality-based weighting.

The patent changes the parameter of weight assignment from uniform (equal weight for all repetitions) to variable (weight based on quality metric). The quality metric serves as a new parameter that reflects the degree of noise disturbance in each repetition. By incorporating this parameter into the decoding process through weighted summation, the method achieves both computational efficiency and accurate differentiation between noise-disturbed and noise-spared signals.

The patent applies preliminary action by pre-defining quality metrics and weighting schemes before the actual decoding process. The method prepares the framework for differentiated evaluation in advance, allowing the system to process repetitions with varying noise levels more effectively. This preliminary preparation maintains implementation simplicity while significantly enhancing error correction capacity through the pre-established quality-based differentiation mechanism.