calibrate_harmonicpower: Quantitative calibration, through harmonic power, for Out and Opn objects

Description

Estimates the number of harmonics required for the four Fourier methods implemented in Momocs: elliptical Fourier analysis (see efourier), radii variation analysis (see rfourier) and tangent angle analysis (see tfourier) and discrete Fourier transform (see dfourier). It returns and can plot cumulated harmonic power whether dropping the first harmonic or not, and based and the maximum possible number of harmonics on the Coo object.

Usage

calibrate_harmonicpower(x, method, id, nb.h, drop, thresh, plot, verbose)

Arguments

a Coo of Opn object

method

any method from c('efourier', 'rfourier', 'sfourier', 'tfourier') for Outs and dfourier for Outs.

the shapes on which to perform calibrate_harmonicpower. All of them by default

nb.h

numeric the maximum number of harmonic, on which to base the cumsum

drop

numeric the number of harmonics to drop for the cumulative sum

thresh

vector of numeric for drawing horizontal lines, and also used for minh below

plot

logical whether to plot the result or simply return the matrix

verbose

whether to print results

Value

returns a list with component:

gg a ggplot object, q the quantile matrix
minh a quick summary that returns the number of harmonics required to achieve a certain proportion of the total harmonic power.

Details

The power of a given harmonic

n

is calculated as follows for elliptical Fourier analysis and the n-th harmonic:

H a r m o n i c P o w e r_{n} \frac{A_{n}^{2} + B_{n}^{2} + C_{n}^{2} + D_{n}^{2}}{2}

and as follows for radii variation and tangent angle:

H a r m o n i c P o w e r_{n} = \frac{A_{n}^{2} + B_{n}^{2} + C_{n}^{2} + D_{n}^{2}}{2}

Examples

Run this code

data(bot)
cal <- calibrate_harmonicpower(bot)
## Not run: ------------------------------------
# # for Opn objects
# data(olea)
# calibrate_harmonicpower(olea, "dfourier")
# 
# # let customize the ggplot
# library(ggplot2)
# cal$gg + theme_minimal() +
# coord_cartesian(xlim=c(3.5, 12.5), ylim=c(90, 100)) +
# ggtitle("Harmonic power calibration")
## ---------------------------------------------
# if you want to do efourier with 99% calibrate_harmonicpower in one step
# efourier(bot, nb.h=calibrate_harmonicpower(bot, "efourier", plot=FALSE)$minh["99%"])

Run the code above in your browser using DataLab

Data engineering and BI courses are free!