ref: fix examples (#2)

examples/3D_example.py

@@ -1,40 +1,31 @@
-#!/usr/bin/env python
-# -*- coding: utf-8 -*-
-""" Example of handling 3D data.
+"""Volumetric data reconstruction
+
 
-Using the `*_3d` methods only speeds up calculations if your processor
-has multiple cores.
 """
-from __future__ import print_function, division
 from multiprocessing import cpu_count
-import numpy as np
-from os.path import split, dirname, abspath
 import time
-import sys
 
-sys.path = [split(dirname(abspath(__file__)))[0]] + sys.path
+import numpy as np
 
 import radontea as rt
 
 
-if __name__ == "__main__":
-    A = 70    # number of angles
-    N = 128   # detector size x
-    M = 24    # detector size y (number of slices)
+A = 70    # number of angles
+N = 128   # detector size x
+M = 24    # detector size y (number of slices)
 
-    # generate random data
-    sino0 = np.random.random((A, N))     # for 2d example
-    sino = np.random.random((A, M, N))    # for 3d example
-    sino[:, 0, :] = sino0
-    angles = np.linspace(0, np.pi, A)     # for both
+# generate random data
+sino0 = np.random.random((A, N))     # for 2d example
+sino = np.random.random((A, M, N))    # for 3d example
+sino[:, 0, :] = sino0
+angles = np.linspace(0, np.pi, A)     # for both
 
-    a = time.time()
-    data0 = rt.backproject(sino0, angles)
-    print("time on 1 core: {} s".format((time.time() - a)*M))
+a = time.time()
+data1 = rt.backproject_3d(sino, angles, ncpus=1)
+print("time on 1 core:  {:.2f} s".format(time.time() - a))
 
-    a = time.time()
-    data = rt._Back_3D.backproject(sino, angles)
-    print("time on {} cores: {} s".format(cpu_count(), time.time() - a))
+a = time.time()
+data2 = rt.backproject_3d(sino, angles, ncpus=cpu_count())
+print("time on {} cores: {:.2f} s".format(cpu_count(), time.time() - a))
 
-    assert np.sum(data0 == data[:, 0, :]
-                  ) == N**2, "2D and 3D results don't match"
+assert np.all(data1 == data2), "2D and 3D results don't match"

examples/comparison_parallel.py

@@ -1,7 +1,7 @@
 """Comparison of parallel-beam reconstruction methods
 
 This example illustrates the performance of the
-different reconstruction techniques for a parralel-beam
+different reconstruction techniques for a parallel-beam
 geometry. The left column shows the reconstruction of
 the original image and the right column shows the reconstruction
 of the corresponding binary images. Note that the
@@ -23,14 +23,14 @@
 angles = np.linspace(0, np.pi, A)
 
 im = get_original(N)
-sino = radontea.radon(im, angles)
+sino = radontea.radon_parallel(im, angles)
 fbp = radontea.backproject(sino, angles)
 fintp = radontea.fourier_map(sino, angles).real
 sarta = radontea.sart(sino, angles, iterations=ITA)
 sartb = radontea.sart(sino, angles, iterations=ITB)
 
 im2 = (im >= (im.max() / 5)) * 255
-sino2 = radontea.radon(im2, angles)
+sino2 = radontea.radon_parallel(im2, angles)
 fbp2 = radontea.backproject(sino2, angles)
 fintp2 = radontea.fourier_map(sino2, angles).real
 sarta2 = radontea.sart(sino2, angles, iterations=ITA)