Hi Luca and others,
I just extended the documentation of mirstat, as follows:
If the object is frame-decomposed, the fields of the output are:
Mean: the average along frames;
Std: the standard deviation along frames;
Slope: the linear slope of the trend along frames, i.e. the
derivative of
the line that would best fit the curve. The slope S is computed in a
normalised representation of the curve C – i.e., centered, with unit
variance, and with a temporal scale reduced to a series T of values
between
0 and 1 – as a solution in a least-square sense of the equation S*T
= C;
PeriodFreq: the frequency (in Hz.) of the maximal periodicity
detected in
the frame-by-frame evolution of the values, estimated through the
computation of the autocorrelation sequence. If no periodicity is
detected,
NaN is returned;
PeriodAmp: the normalized amplitude of that main periodicity, i.e.,
such
that the autocorrelation at zero lag is identically 1.0. The first
descending slope starting from zero lag is removed from this
analysis, as it
is not related to the periodicity of the curve.
PeriodEntropy: the Shannon entropy of the autocorrelation function
(cf.
mirentropy).
Is it clear? Does anyone know any references related to these
concepts?
Olivier
Luca Piccina kirjoitti 11.3.2009 kello 14.21:
Hello,
I am Luca from Italy and I would need some theorical references for
the statistics implemented with mirstat. It's ok with the mean and
the
variance, but I'd like a better understanding of Slope, PeriodFreq,
PeriodAmp and PeriodFreq...could you please give me some references?
Thanks in advance, best wishes
Luca
