[python-dtcwt] 267/497: remote image-registration-1.py example

[Ghislain Vaillant]

This is an automated email from the git hooks/post-receive script.

ghisvail-guest pushed a commit to branch debian/sid
in repository python-dtcwt.

commit 35cd5c05ce0dc481f8d6ee2d7c80447d4368c97e
Author: Rich Wareham <rjw57 at cam.ac.uk>
Date:   Thu Jan 30 11:45:44 2014 +0000

    remote image-registration-1.py example
    
    The code in image-registration-1.py was no longer exemplary of
    best-practice. Remove it rather than letting it bit rot.
---
 docs/examples.rst                                  |   8 +-
 examples/image-registration-1.py                   | 145 ---------------------
 ...age-registration-2.py => image-registration.py} |   0
 3 files changed, 1 insertion(+), 152 deletions(-)

diff --git a/docs/examples.rst b/docs/examples.rst
index 17f8afa..8d444c2 100644
--- a/docs/examples.rst
+++ b/docs/examples.rst
@@ -42,10 +42,4 @@ traffic at a road junction.
 
 The source for the script is shown here:
 
-.. literalinclude:: ../examples/image-registration-2.py
-
-If you're interested in some of the more fundamental theory behind the
-registration, a simpler example which uses lower-level functions to estimate a
-single global affine distortion is available in the
-:download:`image-registration-1.py<../examples/image-registration-1.py>`
-script.
+.. literalinclude:: ../examples/image-registration.py
diff --git a/examples/image-registration-1.py b/examples/image-registration-1.py
deleted file mode 100755
index 1c7da49..0000000
--- a/examples/image-registration-1.py
+++ /dev/null
@@ -1,145 +0,0 @@
-#!/usr/bin/env python
-"""
-An example of image registration via the DTCWT.
-
-This script demonstrates some methods for image registration using the DTCWT.
-
-"""
-
-from __future__ import division, print_function
-
-import os
-import logging
-
-from matplotlib.pyplot import *
-import numpy as np
-
-import dtcwt
-import dtcwt.sampling
-from dtcwt.registration import *
-
-logging.basicConfig(level=logging.INFO)
-
-# Load test image
-logging.info('Loading Lena image')
-f1 = np.load(os.path.join(os.path.dirname(__file__), '..', 'tests', 'lena.npz'))['lena']
-
-# Compute transformed image. Note that we use normalised image co-ords so that the image is on the region [0,1) x [0,1).
-
-logging.info('Creating transformed image')
-xs, ys = np.meshgrid(np.arange(f1.shape[1], dtype=np.float32)/f1.shape[1],
-                     np.arange(f1.shape[0], dtype=np.float32)/f1.shape[0])
-
-def trans(dx, dy):
-    return np.array([[1,0,dx], [0,1,dy], [0,0,1]])
-
-def rot(theta):
-    s, c = np.sin(theta), np.cos(theta)
-    return np.array([[c,s,0], [-s,c,0], [0,0,1]])
-
-theta = np.deg2rad(10)
-dx, dy = 20/f1.shape[1], -30/f1.shape[0]
-
-T = trans(0.5+dx, 0.5+dy).dot(rot(theta)).dot(trans(-0.5,-0.5))
-
-xs_prime = T[0,0]*xs + T[0,1]*ys + T[0,2]
-ys_prime = T[1,0]*xs + T[1,1]*ys + T[1,2]
-
-f2 = dtcwt.sampling.sample(f1, xs_prime*f1.shape[1], ys_prime*f1.shape[0], method='bilinear')
-
-# Also compute a ground truth velocity field. If a point $(x,y)$ is transformed
-# to $(x', y')$ where $[x', y']^T = T \, [x, y]^T$ then:
-#
-# $$ \Delta_x = (x'-x) = T_{11}x + T_{12}y + T_{13} - x $$
-# $$ \Delta_y = (y'-y) = T_{21}x + T_{22}y + T_{23} - y $$
-
-logging.info('Computing ground truth velocity field')
-pxs, pys = np.arange(0, f1.shape[1], 10, dtype=np.float32), np.arange(0, f1.shape[0], 10, dtype=np.float32)
-X, Y = np.meshgrid(pxs, pys)
-X /= f1.shape[1]
-Y /= f1.shape[0]
-gt_vxs = T[0,2] + T[0,0]*X + T[0,1]*Y - X
-gt_vys = T[1,2] + T[1,0]*X + T[1,1]*Y - Y
-
-# Take the DTCWT of both frames.
-logging.info('Taking DTCWT')
-nlevels = 8
-Yl1, Yh1 = dtcwt.dtwavexfm2(f1, nlevels=nlevels)
-Yl2, Yh2 = dtcwt.dtwavexfm2(f2, nlevels=nlevels)
-
-# Iteratively solve for transform
-
-# Start with a set of warped high-pass subbands which
-# are a shallow copy of Yh1
-Yh3 = list(Yh1)
-
-# Initial velocity field is zero
-a = np.zeros(6, dtype=np.float32)
-
-for iteration in xrange(4*(len(Yh1)-5)):
-    startlevel = len(Yh1)-2-iteration//4
-    levels = [startlevel, startlevel+1]
-
-    logging.info('Refining with levels: %s' % (levels,))
-
-    # Warp Yh1 for each level we're looking at in this iteration
-    for level in levels:
-        Yh3[level] = affinewarphighpass(Yh1[level], a, method='bilinear')
-
-    Qt_mats = qtildematrices(Yh3, Yh2, levels)
-    Qt = np.sum(list(np.sum(np.sum(x, axis=0), axis=0) for x in Qt_mats), axis=0)
-    a += solvetransform(Qt)
-
-logging.info('Computing velocity field')
-
-# Compute and plot the reconstructed velocity field and compare it to the ground truth.
-vxs = a[0] + a[2]*X + a[4]*Y
-vys = a[1] + a[3]*X + a[5]*Y
-
-figure()
-
-subplot(121)
-imshow(f1, cmap=cm.gray)
-quiver(pxs, pys, gt_vxs, gt_vys, color='r', angles='xy')
-title('Ground truth velocity field')
-
-subplot(122)
-imshow(f1, cmap=cm.gray)
-quiver(pxs, pys, vxs, vys, color='r', angles='xy')
-title('Computed velocity field')
-
-figure()
-subplot(221)
-imshow(vxs)
-title('Computed velocity x-component')
-colorbar()
-
-subplot(222)
-imshow(gt_vxs)
-title('Ground truth velocity x-component')
-colorbar()
-
-subplot(223)
-imshow(vys)
-title('Computed velocity y-component')
-colorbar()
-
-subplot(224)
-imshow(gt_vys)
-title('Ground truth velocity y-component')
-colorbar()
-
-# Compute a high-quality warping using Lanczos re-sampling
-logging.info('Computing high-quality warped image')
-warped_direct = affinewarp(f1, a, method='lanczos')
-
-figure()
-subplot(121)
-imshow(f2, cmap=cm.gray, clim=(0,1))
-title('Frame 2')
-
-subplot(122)
-imshow(warped_direct, cmap=cm.gray, clim=(0,1))
-title('Frame 1 warped to Frame 2')
-
-show()
diff --git a/examples/image-registration-2.py b/examples/image-registration.py
similarity index 100%
rename from examples/image-registration-2.py
rename to examples/image-registration.py

-- 
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/debian-science/packages/python-dtcwt.git