Simple Peak Detector

used for identifying peaks in time-series experimental data

The detect_peaks algorithm uses rising and falling counters to assign a risingState with user-provided thresholds. Rising count is independent of falling count. A peak is detected when enough rising counts, then falling counts have successfully been detected.

The detect_peaks_with_rise algorithm is the same as detect_peaks except an additional constraint is added before detection. The total rise from the previous peak must be greater than a certain percentage of the total data range. This reduces chances of multiple unwanted repeated peaks. Drops in the data since the last detected peak is incorporated.

A simple Infinite Impulse Repsonse (IIR) filter is included and helpful with noisy data but not always necessary.

Four experimental datasets are provided that illlustrate both algorithms:

1. sinuisoidal, or periodic
2. constant but noisy data with 3 successive peaks
3. rising, stair-step sequence A
4. rising, stair-step sequence B