How to do phase correlation to estimate the geometrical shifts between two images?
Phase Correlation¶
In this notebook I will show how to use Fourier Transform to find the shifts between two misaligned images, then I will show how to use Bi-linear interpolation to align them.¶
In [1]:
import numpy as np
import matplotlib.pyplot as plt
import pyfits
Let's load a continuum image taken from the MURAM simulation of a plage region.¶
In [3]:
path = '/home/fatima/Desktop/solar_orbiter_project/codes/targets/distortion/'
sim = pyfits.getdata(path+'synth_continuum_profile.fits')
In [8]:
fig = plt.figure(figsize=(15,10))
plt.imshow(sim,cmap='gray')
plt.colorbar()
Out[8]:
Let us introduce vertical and horizontal shifts.¶
In [42]:
def shift(x,y,del_x,del_y):
x_s = x + del_x
y_s = y + del_y
return x_s, y_s
In [5]:
from scipy import interpolate
In [12]:
size = sim.shape[0]
x = y = np.arange(size)
F= interpolate.interp2d(x,y,sim, kind='linear')
In [15]:
shifted = np.zeros((size,size))
for j in range(size):
for i in range(size):
i_s, j_s = shift(i,j,10,3)
shifted[j,i] = F(i_s,j_s)
In [16]:
fig=plt.figure(figsize=(25,15))
ax1 = fig.add_subplot(121)
ax2 = fig.add_subplot(122)
ax1.imshow(sim,origin='lower')
ax2.imshow(shifted,origin='lower')
Out[16]: