# -*- coding: utf-8 -*-
"""
Resampling
==========
Module with routines for data and curve resampling and smoothing based on splines.
"""
__author__ = 'Christoph Kirst <christoph.kirst.ck@gmail.com>'
__license__ = 'GPLv3 - GNU General Pulic License v3 (see LICENSE)'
__copyright__ = 'Copyright © 2020 by Christoph Kirst'
__webpage__ = 'http://idisco.info'
__download__ = 'http://www.github.com/ChristophKirst/ClearMap2'
import numpy as np
from scipy.interpolate import splprep,splrep, splev
[docs]def resample_nd(curve, n_points = None, smooth = 0, periodic = False, derivative = 0, order = 5, iterations = 1):
"""Resample n points using n equidistant points along a curve
Arguments
---------
curve : mxd array
Coordinates for the reference points of the curve.
n_points : int
Number of equidistant points to evaluate the curve on.
smooth : int or float
Smoothness factor.
periodic : bool
If True, assume the curve is a closed loop.
derivative : int
If > 0, return the n-th derivative of the curve.
order : int
The spline order to use to interpolate the curve.
iterations : int
Iterate the resampling process this amount of time.
Returns
-------
curve : nxd array
Resampled curve along n_points equidistant points.
"""
if n_points is None or n_points is all:
n_points = curve.shape[0];
for i in range(iterations):
cinterp, u = splprep(curve.T, u=None, s=smooth, per=periodic, k=order);
us = np.linspace(u.min(), u.max(), n_points)
curve = np.vstack(splev(us, cinterp, der=derivative)).T;
return curve;
[docs]def resample_1d(data, n_points = None, smooth = 0, periodic = False, derivative = 0, order = 5, iterations = 0):
"""Resample 1d data using n equidistant points
Arguments
---------
curve : mxd array
Coordinates for the reference points of the curve.
n_points : int
Number of equidistant points to evaluate the curve on.
smooth : int or float
Smoothness factor.
periodic : bool
If True, assume the curve is a closed loop.
derivative : int
If > 0, return the n-th derivative of the curve.
order : int
The spline order to use to interpolate the curve.
iterations : int
Iterate the resampling process this amount of time.
Returns
-------
curve : nxd array
Resampled curve along n_points equidistant points.
"""
if n_points is None or n_points is None:
n_points = data.shape[0];
u0 = np.linspace(0, 1, data.shape[0]);
us = np.linspace(0, 1, n_points);
for i in range(iterations):
dinterp = splrep(u0, data, s=smooth, per=periodic, k=order);
data = splev(us, dinterp, der = derivative);
return data;
[docs]def resample(curve, n_points = None, smooth = 0, periodic = False, derivative = 0, order = 5, iterations = 1):
"""Resample a curve using equidistant points along a curve.
Arguments
---------
curve : mxd array
Coordinates for the reference points of the curve.
n_points : int
Number of equidistant points to evaluate the curve on.
smooth : int or float
Smoothness factor.
periodic : bool
If True, assume the curve is a closed loop.
derivative : int
If > 0, return the n-th derivative of the curve.
order : int
The spline order to use to interpolate the curve.
iterations : int
Iterate the resampling process this amount of time.
Returns
-------
curve : nxd array
Resampled curve along n_points equidistant points.
"""
if curve.ndim > 1:
return resample_nd(curve, n_points, smooth=smooth, periodic=periodic, derivative=derivative, order=order, iterations=iterations);
else:
return resample_1d(curve, n_points, smooth=smooth, periodic=periodic, derivative=derivative, order=order, iterations=iterations);
[docs]def test():
import numpy as np
import matplotlib.pyplot as plt
import ClearMap.Analysis.Curves.Resampling as res
#reload(res)
curve = np.linspace(0,10,50);
curve = np.vstack([curve, np.sin(curve)]).T;
rcurve = res.resample(curve, n_points=150, smooth=0);
plt.figure(1); plt.clf();
plt.plot(rcurve[:,0], rcurve[:,1], 'red');
plt.plot(curve[:,0], curve[:,1], 'blue');
curve1d = np.sin(np.linspace(0,1,50) * 2 * np.pi);
rcurve1d = res.resample(curve1d, n_points=150, smooth=0);
plt.figure(2); plt.clf();
plt.plot(curve1d);
plt.plot(rcurve1d);
Source
