Matplotlib: Colorize grayscale image using another image

Question

I have a black and white spectrum that I want to colorize using this colorize image. colors

Adapting the method given here: Applying different color map to mask, I obtain a final colorized image, but it lacks the features of the spectrum (see comments in code for a link to the picture finalimage).

My code is given here (with the hosted images in the comments since I cannot display all of them here):

import numpy as np
import matplotlib.pyplot as plt
import pyfits

wavemap = "FITS/wavemap.fits.gz"
spectrum = "FITS/phxspectra.fits.gz"

image = pyfits.getdata(spectrum)
colors = pyfits.getdata(wavemap)
mask = colors > 0
colors_ma = np.ma.array(colors, mask=~mask)

kwargs = {'interpolation': 'none', 'vmin': colors[mask].min(), 'vmax': colors.max(), 'origin': 'lower', 'alpha' : 0.5}

plt.imshow(image, cmap=plt.cm.Greys_r, interpolation='none', origin='lower')
plt.imshow(colors_ma, cmap=plt.cm.jet, **kwargs)

plt.show()

#![spectrum](https://i.sstatic.net/NP5mv.png)
#![spectrumfeatures](http://i.imgur.com/MTQ9yMl.png)
#![colorize](https://i.sstatic.net/sOiPT.png)
#![finalimage](http://i.imgur.com/MmnM9qK)
#![finalimagefeatures](http://i.imgur.com/t5PoJiE.png)

If I lower the alpha value, the features of the spectrum show better, but the colors are very dim. If I increase the alpha value, then the colors show much better, but the features of the spectrum do not show.

How can I get the features of the spectrum AND the colors from the colorize image without trading off one for the other?

I don't understand what you are trying to achieve....What are these arrays physically (I assume some sort of astro data)? Is the spectrum array just a vertical gradient sampled along the streaks? — tacaswell
– tacaswell, Commented Oct 9, 2013 at 17:36
I'm sorry, I realize the naming convention is very confusing. Physically, the arrays are 2d spectrums. For the spectrum array, each value is a luminance or intensity value. For the colorize array, coordinate values give the wavelength, or color, of the image, which results in the rainbow gradient. — Manila Thrilla
– Manila Thrilla, Commented Oct 14, 2013 at 7:50
What are the axes? wave length and time? k and omega? what does the white in the second image you posted mean? — tacaswell
– tacaswell, Commented Oct 14, 2013 at 14:22

ali_m · Accepted Answer · 2013-10-13 00:14:35Z

One way would be to create an RGBA array, where the RGB values represent the spectral data, and the A values represent the intensity. Alternatively you could use a HSV colormap, and represent spectral and intensity values using hue and brightness/saturation respectively.

import numpy as np
from matplotlib import pyplot as pp
from scipy.misc import lena

# some 'intensity' data, normalized between 0 and 1
intensity = lena().astype(np.float32)
intensity = (intensity - intensity.min()) / (intensity.max() - intensity.min())

# some 'spectrum' data, normalized between 0 and 1
x,y = np.mgrid[-1:1:intensity.shape[0]*1j,-1:1:intensity.shape[1]*1j]
spectrum = 0.5*(np.sin(20*(x**2 + y**2)) + 1)

# look up RGB values from the 'jet' colormap
RGBA = pp.cm.jet(spectrum)
# fill the A(lpha) channel with the array of intensity data
RGBA[...,3] = intensity

# another way to represent spectral and intensity data is using a 2D 
# HSV color space
HSV = np.ones(intensity.shape + (3,))

# we represent the spectral information using hue...
HSV[...,0] = spectrum
# ... and the intensity data as brightness and saturation
HSV[...,1] = intensity
HSV[...,2] = intensity
# convert back into rgb color space
HSV = matplotlib.colors.hsv_to_rgb(HSV)

# plotting
fig, ax = pp.subplots(2, 2, figsize=(10,10), sharex=True, sharey=True)
ax[0,0].imshow(intensity, cmap=pp.cm.gray)
ax[0,0].set_title('Intensity data', fontsize=14)
ax[0,1].imshow(spectrum, cmap=pp.cm.gray)
ax[0,1].set_title('Spectral data', fontsize=14)
ax[1,0].imshow(RGBA)
ax[1,0].set_title('Jet with varying alpha', fontsize=14)
ax[1,1].imshow(HSV)
ax[1,1].set_title('HSV', fontsize=14)
for aa in ax.flat:
    aa.set_axis_bgcolor((0,0,0,1))
    aa.xaxis.set_visible(False)
    aa.yaxis.set_visible(False)
fig.tight_layout()

pp.show()

enter image description here

Just what I was looking for! However, I ended up using HSV[...,2] = np.ones(intensity.size)

Collectives™ on Stack Overflow

Matplotlib: Colorize grayscale image using another image

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related