Utility class that looks for a peak in data looking at the data one point at a time.
Note: I wrote this for finding peaks on well behaved sinusoidal data. I have some Ideas in the comments if the data is not as well behaved.
| biogears::DataPeak::DataPeak |
( |
unsigned int |
samples, |
|
|
double |
epsilon |
|
) |
| |
– Number of Samples - This is how many values the peak is compared too, that the current peak must be greater than. When the this many samples are smaller than the peak, we have found a peak. – epsilon - The window (below the peak if trending up, above the peak if trending down) that a new value must be below/above in order to be considered treding away from the peak If data is inside the window, we are in a plataue and don't do anything. If we are going away from the peak (opposite direction) we need #samples away from the window to signal a peak. If we find a point that is a new peak, we move our window and start our counting all over.
By default this class will look for max peak only By default this class assumes data coming in is trening in the positive/up direction
