Tradeoff Analysis of Discrepancy-Based Adaptive Thresholding Approach

Weyl's discrepancy measure distinguishes by its property to best-approximate isometry for threshold based sampling. We discuss how the resulting quasi-isometry motivates the design of novel adaptive thresholding approaches. Our experimental analysis on the basis of send-on-delta samples shows that a significant reduction in the number of samples can be achieved with an approximately unchanged signal-to-noise ratio.