I'm trying to plot the Amplitude (dBFS) vs. Time (s) plot of an audio (.wav) file using matplotlib. I managed to do that with the following code:
def convert_to_decibel(sample):
ref = 32768 # Using a signed 16-bit PCM format wav file. So, 2^16 is the max. value.
if sample!=0:
return 20 * np.log10(abs(sample) / ref)
else:
return 20 * np.log10(0.000001)
from scipy.io.wavfile import read as readWav
from scipy.fftpack import fft
import matplotlib.pyplot as gplot1
import matplotlib.pyplot as gplot2
import numpy as np
import struct
import gc
wavfile1 = '/home/user01/audio/speech.wav'
wavsamplerate1, wavdata1 = readWav(wavfile1)
wavdlen1 = wavdata1.size
wavdtype1 = wavdata1.dtype
gplot1.rcParams['figure.figsize'] = [15, 5]
pltaxis1 = gplot1.gca()
gplot1.axhline(y=0, c="black")
gplot1.xticks(np.arange(0, 10, 0.5))
gplot1.yticks(np.arange(-200, 200, 5))
gplot1.grid(linestyle = '--')
wavdata3 = np.array([convert_to_decibel(i) for i in wavdata1], dtype=np.int16)
yvals3 = wavdata3
t3 = wavdata3.size / wavsamplerate1
xvals3 = np.linspace(0, t3, wavdata3.size)
pltaxis1.set_xlim([0, t3 + 2])
pltaxis1.set_title('Amplitude (dBFS) vs Time(s)')
pltaxis1.plot(xvals3, yvals3, '-')
which gives the following output:
I had also plotted the Power Spectral Density (PSD, in dBm) using the code below:
from scipy.signal import welch as psd # Computes PSD using Welch's method.
fpsd, wPSD = psd(wavdata1, wavsamplerate1, nperseg=1024)
gplot2.rcParams['figure.figsize'] = [15, 5]
pltpsdm = gplot2.gca()
gplot2.axhline(y=0, c="black")
pltpsdm.plot(fpsd, 20*np.log10(wPSD))
gplot2.xticks(np.arange(0, 4000, 400))
gplot2.yticks(np.arange(-150, 160, 10))
pltpsdm.set_xlim([0, 4000])
pltpsdm.set_ylim([-150, 150])
gplot2.grid(linestyle = '--')
which gives the output as:
The second output above, using the Welch's method plots a more presentable output. The dBFS plot though informative is not very presentable IMO. Is this because of:
the difference in the domains (time in case of 1st output vs frequency in the 2nd output)?
the way plot function is implemented in pyplot?
Also, is there a way I can plot my dBFS output as a peak-to-peak style of plot just like in my PSD (dBm) plot rather than a dense stem plot?
Would be much helpful and would appreciate any pointers, answers or suggestions from experts here as I'm just a beginner with matplotlib and plots in python in general.
TLNR
This has nothing to do with pyplot.
The frequency domain is different from the time domain, but that's not why you didn't get what you want.
The calculation of dbFS in your code is wrong.
You should frame your data, calculate RMSs or peaks in every frame, and then convert that value to dbFS instead of applying this transformation to every sample point.
When we talk about the amplitude, we are talking about a periodic signal. And when we read in a series of data from a sound file, we read in a series of sample points of a signal(may be or be not periodic). The value of every sample point represents a, say, voltage value, or sound pressure value sampled at a specific time.
We assume that, within a very short time interval, maybe 10ms for example, the signal is stationary. Every such interval is called a frame.
Some specific function is applied to each frame usually, to reduce the sudden change at the edge of this frame, and these functions are called window functions. If you did nothing to every frame, you added rectangle windows to them.
An example: when the sampling frequency of your sound is 44100Hz, in a 10ms-long frame, there are 44100*0.01=441 sample points. That's what the nperseg argument means in your psd function but it has nothing to do with dbFS.
Given the knowledge above, now we can talk about the amplitude.
There are two methods a get the value of amplitude in every frame:
The most straightforward one is to get the maximum(peak) values in every frame.
Another one is to calculate the RMS(Root Mean Sqaure) of every frame.
After that, the peak values or RMS values can be converted to dbFS values.
Let's start coding:
import numpy as np
import matplotlib.pyplot as plt
from scipy.io import wavfile
# Determine full scall(maximum possible amplitude) by bit depth
bit_depth = 16
full_scale = 2 ** bit_depth
# dbFS function
to_dbFS = lambda x: 20 * np.log10(x / full_scale)
# Read in the wave file
fname = "01.wav"
fs,data = wavfile.read(fname)
# Determine frame length(number of sample points in a frame) and total frame numbers by window length(how long is a frame in seconds)
window_length = 0.01
signal_length = data.shape[0]
frame_length = int(window_length * fs)
nframes = signal_length // frame_length
# Get frames by broadcast. No overlaps are used.
idx = frame_length * np.arange(nframes)[:,None] + np.arange(frame_length)
frames = data[idx].astype("int64") # Convert to in 64 to avoid integer overflow
# Get RMS and peaks
rms = ((frames**2).sum(axis=1)/frame_length)**.5
peaks = np.abs(frames).max(axis=1)
# Convert them to dbfs
dbfs_rms = to_dbFS(rms)
dbfs_peak = to_dbFS(peaks)
# Let's start to plot
# Get time arrays of every sample point and ever frame
frame_time = np.arange(nframes) * window_length
data_time = np.linspace(0,signal_length/fs,signal_length)
# Plot
f,ax = plt.subplots()
ax.plot(data_time,data,color="k",alpha=.3)
# Plot the dbfs values on a twin x Axes since the y limits are not comparable between data values and dbfs
tax = ax.twinx()
tax.plot(frame_time,dbfs_rms,label="RMS")
tax.plot(frame_time,dbfs_peak,label="Peak")
tax.legend()
f.tight_layout()
# Save serval details
f.savefig("whole.png",dpi=300)
ax.set_xlim(1,2)
f.savefig("1-2sec.png",dpi=300)
ax.set_xlim(1.295,1.325)
f.savefig("1.2-1.3sec.png",dpi=300)
The whole time span looks like(the unit of the right axis is dbFS):
And the voiced part looks like:
You can see that the dbFS values become greater while the amplitudes become greater at the vowel start point:
I'm using openturns to find the best fit distribution for my data. I got to plot it alright, but the X limit is far bigger than I'd like. My code is:
import statsmodels.api as sm
import openturns as ot
import openturns.viewer as otv
data = in_seconds
sample = ot.Sample(data, 1)
tested_factories = ot.DistributionFactory.GetContinuousUniVariateFactories()
best_model, best_bic = ot.FittingTest.BestModelBIC(sample, tested_factories)
print(best_model)
graph = ot.HistogramFactory().build(sample).drawPDF()
bestPDF = best_model.drawPDF()
bestPDF.setColors(["blue"])
graph.add(bestPDF)
name = best_model.getImplementation().getClassName()
graph.setLegends(["Histogram",name])
graph.setXTitle("Latências (segundos)")
graph.setYTitle("Frequência")
otv.View(graph)
I'd like to set X limits as something like "graph.setXLim", as we'd do in matplotlib, but I'm stuck with it as I'm new to OpenTurns.
Thanks in advance.
Any OpenTURNS graph has a getBoundingBox method which returns the bounding box as a dimension 2 Interval. We can get/set the lower and upper bounds of this interval with getLowerBound and getUpperBound. Each of these bounds is a Point with dimension 2. Hence, we can set the bounds of the graphics prior to the use of the View class.
In the following example, I create a simple graph containing the PDF of the gaussian distribution.
import openturns as ot
import openturns.viewer as otv
n = ot.Normal()
graph = n.drawPDF()
_ = otv.View(graph)
Suppose that I want to set the lower X axis to -1.
The script:
boundingBox = graph.getBoundingBox()
lb = boundingBox.getLowerBound()
print(lb)
produces:
[-4.10428,-0.0195499]
The first value in the Point is the X lower bound and the second is the Y lower bound. The following script sets the first component of the lower bound to -1, wraps the lower bound into the bounding box and sets the bounding box into the graph.
lb[0] = -1.0
boundingBox.setLowerBound(lb)
graph.setBoundingBox(boundingBox)
_ = otv.View(graph)
This produces the following graph.
The advantage of these methods is that they configure the graphics from the library, before the rendering is done by Matplotlib. The drawback is that they are a little more verbose than the Matplotlib counterpart.
Here is a minimal example adapted from openTURNS examples (see http://openturns.github.io/openturns/latest/examples/graphs/graphs_basics.html) to set the x range (initially from [-4,4] to [-2,2]) :
import openturns as ot
import openturns.viewer as viewer
from matplotlib import pylab as plt
n = ot.Normal()
# To configure the look of the plot, we can first observe the type
# of graphics returned by the `drawPDF` method returns: it is a `Graph`.
graph = n.drawPDF()
# The `Graph` class provides several methods to configure the legends,
# the title and the colors. Since a graphics can contain several sub-graphics,
# the `setColors` takes a list of colors as inputs argument: each item of
# the list corresponds to the sub-graphics.
graph.setXTitle("N")
graph.setYTitle("PDF")
graph.setTitle("Probability density function of the standard gaussian distribution")
graph.setLegends(["N"])
graph.setColors(["blue"])
# Combine several graphics
# In order to combine several graphics, we can use the `add` method.
# Let us create an empirical histogram from a sample.
sample = n.getSample(100)
histo = ot.HistogramFactory().build(sample).drawPDF()
# Then we add the histogram to the `graph` with the `add` method.
# The `graph` then contains two plots.
graph.add(histo)
# Using openturns.viewer
view = viewer.View(graph)
# Get the matplotlib.axes.Axes member with getAxes()
# Similarly, there is a getFigure() method as well
axes = view.getAxes() # axes is a matplotlib object
_ = axes[0].set_xlim(-2.0, 2.0)
plt.show()
You can read the definition of the View object here :
https://github.com/openturns/openturns/blob/master/python/src/viewer.py
As you will see, the View class contains matplotlib objects such as axes and figure. Once accessed by the getAxes (or getFigure) you can use the matplotlib methods.
Using PyWavelets and Matplotbib.Specgram on a signal gives more detailed plots with pywt.dwt then pywt.cwt. How can I get a pywt.cwt specgram in a similar way?
With dwt:
import pywt
import pywt.data
import matplotlib.pyplot as plot
from scipy import signal
from scipy.io import wavfile
bA, bD = pywt.dwt(datamean, 'db2')
powerSpectrum, freqenciesFound, time, imageAxis = plot.specgram(bA, NFFT = 387, Fs=100)
plot.xlabel('Time')
plot.ylabel('Frequency')
plot.show()
with this spectrogram plot:
https://imgur.com/a/bYb8bBS
With cwt:
widths = np.arange(1,5)
coef, freqs = pywt.cwt(datamean, widths,'morl')
powerSpectrum, freqenciesFound, time, imageAxis = plot.specgram(coef, NFFT = 129, Fs=100)
plot.xlabel('Time')
plot.ylabel('Frequency')
plot.show()
with this spectrogram plot:
https://imgur.com/a/GIINzJp
and for better results:
sig = datamean
widths = np.arange(1, 31)
cwtmatr = signal.cwt(sig, signal.ricker, widths)
plt.imshow(cwtmatr, extent=[-1, 1, 1, 5], cmap='PRGn', aspect='auto',
vmax=abs(cwtmatr).max(), vmin=-abs(cwtmatr).max())
plt.show()
with this spectrogram plot:
https://imgur.com/a/TnXqgGR
How can I get for cwt (spectrogram plot 2 and 3) a similar spectogram plot and style like in the first one?
It seems like the 1st spectrogram plot compared to the 3rd has much more details.
This would be better as a comment, but since I lack the Karma to do that:
You don't want to make a spectrogram with wavelets, but a scalogram instead. What it looks like you're doing above is projecting your data in a scale subspace (that correlates to frequency), then taking those scales and finding the frequency content of them which is not what you probably want.
The detail and approximation coefficients are what you would want to use directly. Unfortunately, PyWavelets doesn't have a simple plotting function to do this for you, AFAIK. Matlab does, and their help page may be illuminating if I fail.
def scalogram(data):
wave='db4'
coeff=pywt.wavedec(data,wave)
levels=len(coeff)
lengths=[len(co) for co in coeff]
col=np.max(lengths)
im=np.ones([levels,col])
col=col.astype(float)
for level in range(levels):
#print [lengths[level],col]
y=coeff[level]
if lengths[1+level]<col:
x=col/(lengths[1+level]+1)*np.arange(1,len(y)+1)
xi=np.linspace(0,int(col),int(col))
yi=griddata(points=x,values=y,xi=xi,method='nearest')
else:
yi=y
im[level,:]=yi
im[im==0]=np.nan
tiles=sum(lengths)-lengths[0]
return im,tiles
Wxx,tiles=scalogram(data)
IM=plt.imshow(np.log10(abs(Wxx)),aspect='auto')
plt.show()
There are better ways of doing that, but it works. This produces a square matrix similar to spectrogram in "Wxx", and tiles is simply a counter of the number of time-frequency tilings to compare to the number used in a SFFT.
I've attached a picture of what these tilings look like
I am trying to plot normal distribution curve using Python. First I did it manually by using the normal probability density function and then I found there's an exiting function pdf in scipy under stats module. However, the results I get are quite different.
Below is the example that I tried:
import numpy as np
import matplotlib.pyplot as plt
import scipy.stats as stats
mean = 5
std_dev = 2
num_dist = 50
# Draw random samples from a normal (Gaussion) distribution
normalDist_dataset = np.random.normal(mean, std_dev, num_dist)
# Sort these values.
normalDist_dataset = sorted(normalDist_dataset)
# Create the bins and histogram
plt.figure(figsize=(15,7))
count, bins, ignored = plt.hist(normalDist_dataset, num_dist, density=True)
new_mean = np.mean(normalDist_dataset)
new_std = np.std(normalDist_dataset)
normal_curve1 = stats.norm.pdf(normalDist_dataset, new_mean, new_std)
normal_curve2 = (1/(new_std *np.sqrt(2*np.pi))) * (np.exp(-(bins - new_mean)**2 / (2 * new_std**2)))
plt.plot(normalDist_dataset, normal_curve1, linewidth=4, linestyle='dashed')
plt.plot(bins, normal_curve2, linewidth=4, color='y')
The result shows how the two curves I get are very different from each other.
My guess is that it is has something to do with bins or pdf behaves differently than usual formula. I have used the same and new mean and standard deviation for both the plots. So, how do I change my code to match what stats.norm.pdf is doing?
I don't know yet which curve is correct.
Function plot simply connects the dots with line segments. Your bins do not have enough dots to show a smooth curve. Possible solution:
....
normal_curve1 = stats.norm.pdf(normalDist_dataset, new_mean, new_std)
bins = normalDist_dataset # Add this line
normal_curve2 = (1/(new_std *np.sqrt(2*np.pi))) * (np.exp(-(bins - new_mean)**2 / (2 * new_std**2)))
....
I am trying to plot contour lines of pressure level. I am using a netCDF file which contain the higher resolution data (ranges from 3 km to 27 km). Due to higher resolution data set, I get lot of pressure values which are not required to be plotted (rather I don't mind omitting certain contour line of insignificant values). I have written some plotting script based on the examples given in this link http://matplotlib.org/basemap/users/examples.html.
After plotting the image looks like this
From the image I have encircled the contours which are small and not required to be plotted. Also, I would like to plot all the contour lines smoother as mentioned in the above image. Overall I would like to get the contour image like this:-
Possible solution I think of are
Find out the number of points required for plotting contour and mask/omit those lines if they are small in number.
or
Find the area of the contour (as I want to omit only circled contour) and omit/mask those are smaller.
or
Reduce the resolution (only contour) by increasing the distance to 50 km - 100 km.
I am able to successfully get the points using SO thread Python: find contour lines from matplotlib.pyplot.contour()
But I am not able to implement any of the suggested solution above using those points.
Any solution to implement the above suggested solution is really appreciated.
Edit:-
# Andras Deak
I used print 'diameter is ', diameter line just above del(level.get_paths()[kp]) line to check if the code filters out the required diameter. Here is the filterd messages when I set if diameter < 15000::
diameter is 9099.66295612
diameter is 13264.7838257
diameter is 445.574234531
diameter is 1618.74618114
diameter is 1512.58974168
However the resulting image does not have any effect. All look same as posed image above. I am pretty sure that I have saved the figure (after plotting the wind barbs).
Regarding the solution for reducing the resolution, plt.contour(x[::2,::2],y[::2,::2],mslp[::2,::2]) it works. I have to apply some filter to make the curve smooth.
Full working example code for removing lines:-
Here is the example code for your review
#!/usr/bin/env python
from netCDF4 import Dataset
import matplotlib
matplotlib.use('agg')
import matplotlib.pyplot as plt
import numpy as np
import scipy.ndimage
from mpl_toolkits.basemap import interp
from mpl_toolkits.basemap import Basemap
# Set default map
west_lon = 68
east_lon = 93
south_lat = 7
north_lat = 23
nc = Dataset('ncfile.nc')
# Get this variable for later calucation
temps = nc.variables['T2']
time = 0 # We will take only first interval for this example
# Draw basemap
m = Basemap(projection='merc', llcrnrlat=south_lat, urcrnrlat=north_lat,
llcrnrlon=west_lon, urcrnrlon=east_lon, resolution='l')
m.drawcoastlines()
m.drawcountries(linewidth=1.0)
# This sets the standard grid point structure at full resolution
x, y = m(nc.variables['XLONG'][0], nc.variables['XLAT'][0])
# Set figure margins
width = 10
height = 8
plt.figure(figsize=(width, height))
plt.rc("figure.subplot", left=.001)
plt.rc("figure.subplot", right=.999)
plt.rc("figure.subplot", bottom=.001)
plt.rc("figure.subplot", top=.999)
plt.figure(figsize=(width, height), frameon=False)
# Convert Surface Pressure to Mean Sea Level Pressure
stemps = temps[time] + 6.5 * nc.variables['HGT'][time] / 1000.
mslp = nc.variables['PSFC'][time] * np.exp(9.81 / (287.0 * stemps) * nc.variables['HGT'][time]) * 0.01 + (
6.7 * nc.variables['HGT'][time] / 1000)
# Contour only at 2 hpa interval
level = []
for i in range(mslp.min(), mslp.max(), 1):
if i % 2 == 0:
if i >= 1006 and i <= 1018:
level.append(i)
# Save mslp values to upload to SO thread
# np.savetxt('mslp.txt', mslp, fmt='%.14f', delimiter=',')
P = plt.contour(x, y, mslp, V=2, colors='b', linewidths=2, levels=level)
# Solution suggested by Andras Deak
for level in P.collections:
for kp,path in enumerate(level.get_paths()):
# include test for "smallness" of your choice here:
# I'm using a simple estimation for the diameter based on the
# x and y diameter...
verts = path.vertices # (N,2)-shape array of contour line coordinates
diameter = np.max(verts.max(axis=0) - verts.min(axis=0))
if diameter < 15000: # threshold to be refined for your actual dimensions!
#print 'diameter is ', diameter
del(level.get_paths()[kp]) # no remove() for Path objects:(
#level.remove() # This does not work. produces ValueError: list.remove(x): x not in list
plt.gcf().canvas.draw()
plt.savefig('dummy', bbox_inches='tight')
plt.close()
After the plot is saved I get the same image
You can see that the lines are not removed yet. Here is the link to mslp array which we are trying to play with http://www.mediafire.com/download/7vi0mxqoe0y6pm9/mslp.txt
If you want x and y data which are being used in the above code, I can upload for your review.
Smooth line
You code to remove the smaller circles working perfectly. However the other question I have asked in the original post (smooth line) does not seems to work. I have used your code to slice the array to get minimal values and contoured it. I have used the following code to reduce the array size:-
slice = 15
CS = plt.contour(x[::slice,::slice],y[::slice,::slice],mslp[::slice,::slice], colors='b', linewidths=1, levels=levels)
The result is below.
After searching for few hours I found this SO thread having simmilar issue:-
Regridding regular netcdf data
But none of the solution provided over there works.The questions similar to mine above does not have proper solutions. If this issue is solved then the code is perfect and complete.
General idea
Your question seems to have 2 very different halves: one about omitting small contours, and another one about smoothing the contour lines. The latter is simpler, since I can't really think of anything else other than decreasing the resolution of your contour() call, just like you said.
As for removing a few contour lines, here's a solution which is based on directly removing contour lines individually. You have to loop over the collections of the object returned by contour(), and for each element check each Path, and delete the ones you don't need. Redrawing the figure's canvas will get rid of the unnecessary lines:
# dummy example based on matplotlib.pyplot.clabel example:
import matplotlib
import numpy as np
import matplotlib.cm as cm
import matplotlib.mlab as mlab
import matplotlib.pyplot as plt
delta = 0.025
x = np.arange(-3.0, 3.0, delta)
y = np.arange(-2.0, 2.0, delta)
X, Y = np.meshgrid(x, y)
Z1 = mlab.bivariate_normal(X, Y, 1.0, 1.0, 0.0, 0.0)
Z2 = mlab.bivariate_normal(X, Y, 1.5, 0.5, 1, 1)
# difference of Gaussians
Z = 10.0 * (Z2 - Z1)
plt.figure()
CS = plt.contour(X, Y, Z)
for level in CS.collections:
for kp,path in reversed(list(enumerate(level.get_paths()))):
# go in reversed order due to deletions!
# include test for "smallness" of your choice here:
# I'm using a simple estimation for the diameter based on the
# x and y diameter...
verts = path.vertices # (N,2)-shape array of contour line coordinates
diameter = np.max(verts.max(axis=0) - verts.min(axis=0))
if diameter<1: # threshold to be refined for your actual dimensions!
del(level.get_paths()[kp]) # no remove() for Path objects:(
# this might be necessary on interactive sessions: redraw figure
plt.gcf().canvas.draw()
Here's the original(left) and the removed version(right) for a diameter threshold of 1 (note the little piece of the 0 level at the top):
Note that the top little line is removed while the huge cyan one in the middle doesn't, even though both correspond to the same collections element i.e. the same contour level. If we didn't want to allow this, we could've called CS.collections[k].remove(), which would probably be a much safer way of doing the same thing (but it wouldn't allow us to differentiate between multiple lines corresponding to the same contour level).
To show that fiddling around with the cut-off diameter works as expected, here's the result for a threshold of 2:
All in all it seems quite reasonable.
Your actual case
Since you've added your actual data, here's the application to your case. Note that you can directly generate the levels in a single line using np, which will almost give you the same result. The exact same can be achieved in 2 lines (generating an arange, then selecting those that fall between p1 and p2). Also, since you're setting levels in the call to contour, I believe the V=2 part of the function call has no effect.
import numpy as np
import matplotlib.pyplot as plt
# insert actual data here...
Z = np.loadtxt('mslp.txt',delimiter=',')
X,Y = np.meshgrid(np.linspace(0,300000,Z.shape[1]),np.linspace(0,200000,Z.shape[0]))
p1,p2 = 1006,1018
# this is almost the same as the original, although it will produce
# [p1, p1+2, ...] instead of `[Z.min()+n, Z.min()+n+2, ...]`
levels = np.arange(np.maximum(Z.min(),p1),np.minimum(Z.max(),p2),2)
#control
plt.figure()
CS = plt.contour(X, Y, Z, colors='b', linewidths=2, levels=levels)
#modified
plt.figure()
CS = plt.contour(X, Y, Z, colors='b', linewidths=2, levels=levels)
for level in CS.collections:
for kp,path in reversed(list(enumerate(level.get_paths()))):
# go in reversed order due to deletions!
# include test for "smallness" of your choice here:
# I'm using a simple estimation for the diameter based on the
# x and y diameter...
verts = path.vertices # (N,2)-shape array of contour line coordinates
diameter = np.max(verts.max(axis=0) - verts.min(axis=0))
if diameter<15000: # threshold to be refined for your actual dimensions!
del(level.get_paths()[kp]) # no remove() for Path objects:(
# this might be necessary on interactive sessions: redraw figure
plt.gcf().canvas.draw()
plt.show()
Results, original(left) vs new(right):
Smoothing by resampling
I've decided to tackle the smoothing problem as well. All I could come up with is downsampling your original data, then upsampling again using griddata (interpolation). The downsampling part could also be done with interpolation, although the small-scale variation in your input data might make this problem ill-posed. So here's the crude version:
import scipy.interpolate as interp #the new one
# assume you have X,Y,Z,levels defined as before
# start resampling stuff
dN = 10 # use every dN'th element of the gridded input data
my_slice = [slice(None,None,dN),slice(None,None,dN)]
# downsampled data
X2,Y2,Z2 = X[my_slice],Y[my_slice],Z[my_slice]
# same as X2 = X[::dN,::dN] etc.
# upsampling with griddata over original mesh
Zsmooth = interp.griddata(np.array([X2.ravel(),Y2.ravel()]).T,Z2.ravel(),(X,Y),method='cubic')
# plot
plt.figure()
CS = plt.contour(X, Y, Zsmooth, colors='b', linewidths=2, levels=levels)
You can freely play around with the grids used for interpolation, in this case I just used the original mesh, as it was at hand. You can also play around with different kinds of interpolation: the default 'linear' one will be faster, but less smooth.
Result after downsampling(left) and upsampling(right):
Of course you should still apply the small-line-removal algorithm after this resampling business, and keep in mind that this heavily distorts your input data (since if it wasn't distorted, then it wouldn't be smooth). Also, note that due to the crude method used in the downsampling step, we introduce some missing values near the top/right edges of the region under consideraton. If this is a problem, you should consider doing the downsampling based on griddata as I've noted earlier.
This is a pretty bad solution, but it's the only one that I've come up with. Use the get_contour_verts function in this solution you linked to, possibly with the matplotlib._cntr module so that nothing gets plotted initially. That gives you a list of contour lines, sections, vertices, etc. Then you have to go through that list and pop the contours you don't want. You could do this by calculating a minimum diameter, for example; if the max distance between points is less than some cutoff, throw it out.
That leaves you with a list of LineCollection objects. Now if you make a Figure and Axes instance, you can use Axes.add_collection to add all of the LineCollections in the list.
I checked this out really quick, but it seemed to work. I'll come back with a minimum working example if I get a chance. Hope it helps!
Edit: Here's an MWE of the basic idea. I wasn't familiar with plt._cntr.Cntr, so I ended up using plt.contour to get the initial contour object. As a result, you end up making two figures; you just have to close the first one. You can replace checkDiameter with whatever function works. I think you could turn the line segments into a Polygon and calculate areas, but you'd have to figure that out on your own. Let me know if you run into problems with this code, but it at least works for me.
import numpy as np
import matplotlib as mpl
import matplotlib.pyplot as plt
def checkDiameter(seg, tol=.3):
# Function for screening line segments. NB: Not actually a proper diameter.
diam = (seg[:,0].max() - seg[:,0].min(),
seg[:,1].max() - seg[:,1].min())
return not (diam[0] < tol or diam[1] < tol)
# Create testing data
x = np.linspace(-1,1, 21)
xx, yy = np.meshgrid(x,x)
z = np.exp(-(xx**2 + .5*yy**2))
# Original plot with plt.contour
fig0, ax0 = plt.subplots()
# Make sure this contour object actually has a tiny contour to remove
cntrObj = ax0.contour(xx,yy,z, levels=[.2,.4,.6,.8,.9,.95,.99,.999])
# Primary loop: Copy contours into a new LineCollection
lineNew = list()
for lineOriginal in cntrObj.collections:
# Get properties of the original LineCollection
segments = lineOriginal.get_segments()
propDict = lineOriginal.properties()
propDict = {key: value for (key,value) in propDict.items()
if key in ['linewidth','color','linestyle']} # Whatever parameters you want to carry over
# Filter out the lines with small diameters
segments = [seg for seg in segments if checkDiameter(seg)]
# Create new LineCollection out of the OK segments
if len(segments) > 0:
lineNew.append(mpl.collections.LineCollection(segments, **propDict))
# Make new plot with only these line collections; display results
fig1, ax1 = plt.subplots()
ax1.set_xlim(ax0.get_xlim())
ax1.set_ylim(ax0.get_ylim())
for line in lineNew:
ax1.add_collection(line)
plt.show()
FYI: The bit with propDict is just to automate bringing over some of the line properties from the original plot. You can't use the whole dictionary at once, though. First, it contains the old plot's line segments, but you can just swap those for the new ones. But second, it appears to contain a number of parameters that are in conflict with each other: multiple linewidths, facecolors, etc. The {key for key in propDict if I want key} workaround is my way to bypass that, but I'm sure someone else can do it more cleanly.