Source code for freeart.utils.benchmark

# coding: utf-8
# /*##########################################################################
#
# Copyright (c) 2016 European Synchrotron Radiation Facility
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
# THE SOFTWARE.
#
# ###########################################################################*/
"""Benchmark file."""

__authors__ = ["H. Payno"]
__license__ = "MIT"
__date__ = "05/07/2016"

import freeart
from freeart.utils import reconstrutils, exampleutils, outgoingrayalgorithm, raypointsmethod

from silx.gui.plot import Plot2D
from silx.gui import qt

import sys

savingToPdf = False

[docs]def benchmark_tx_projection(oversamplings, widthsToTest, nbAngle): """ Create a graph for benchmarking the transmission forward projector :param oversamplings: the list of oversamplings to test :param widthsToTest: the list of phantom width to test :param nbAngle: the number of projection to test """ times = {} for oversamp in oversamplings : bid = "oversamp=" + str(oversamp) times[bid] = {} for width in widthsToTest : times[bid][width] = exampleutils.produce_tx_SheppLogan_sinogram(oversamp, nbAngle, width, save=False)[2] print("tx forward projection - execution time for oversamp = " + str(oversamp) + ", width = " + str(width) + " toke : " + str(times[bid][width]) + " s") if savingToPdf : title= "tx_benchmatk_for_forward_projection_on_" + str(nbAngle) + "_angles" plot = plot2D(title=title ) for curve in times: plot.addCurve(widthsToTest, times[curve].values(), legend=curve ) plot.legendsDockWidget.show() plot.saveGraph(filename=title + '.png', fileFormat='png') plot.hide()
[docs]def benchmark_tx_reconstruction(oversamplings, widthsToTest, nbAngle, nbIter): """ Create a graph for benchmarking the transmission reconstructor :param oversamplings: the list of oversamplings to test :param widthsToTest: the list of phantom width to test :param nbAngle: the number of projection to test :param nbIter: the number of iteration to test (one iteration launch nbAngle back projection) """ times = {} for oversamp in oversamplings : bid = "oversamp=" + str(oversamp) times[bid] = {} for width in widthsToTest : sino, angles, time = exampleutils.produce_tx_SheppLogan_sinogram(_oversampling=oversamp, _anglesNb=nbAngle, width=width, save=False ) times[bid][width] = exampleutils.make_tx_reconstruction(sino, angles, _nbIter=nbIter)[1] print("tx reconstruction - execution time for oversamp = " + str(oversamp) + ", width = " + str(width) + " toke : " + str(times[bid][width]) + " s") if savingToPdf : title = "tx_benchmatk_for_reconstruction_on_" + str(nbAngle) + "_angles_and_" + str(nbIter) + "_iteration" plot = plot2D(title=title ) for curve in times: plot.addCurve(widthsToTest, times[curve].values(), legend=curve ) plot.legendsDockWidget.show() plot.saveGraph(filename=(title + ".png"), fileFormat="png") plot.hide()
[docs]def benchmark_fluo_projection(oversamplings, widthsToTest, nbAngle, outgoingraypointsmethods): """ Benchmark the fluo projection :param oversamplings: the list of oversamplings to test :param widthsToTest: the list of phantom width to test :param nbAngle: the number of projection to test :param outgoingraypointsmethods: the list of outgoingrayalgorithm to test """ times = {} for oversamp in oversamplings : for obmc in outgoingraypointsmethods: bid = "method="+ str(obmc) + "_oversamp=" + str(oversamp) times[bid] = {} for width in widthsToTest : times[bid][width] = exampleutils.produce_fluo_SheppLogan_sinogram(oversamp, nbAngle, detSetup, width, save=False)[4] print("fluo projection - execution time for " + str(oversamp) + " oversampling, with = " + str(width) + " toke : " + str(times[bid][width])) if savingToPdf : for obmc in outgoingraypointsmethods: title = "fluo_benchmark_for_projection_method_" + str(obmc) + "_on_" + str(nbAngle) + "_angles_and_" plot = plot2D(title=title ) for curve in times: plot.addCurve(widthsToTest, times[curve].values(), legend=curve ) plot.legendsDockWidget.show() plot.saveGraph(filename=(title + ".png"), fileFormat="png") plot.hide()
[docs]def benchmark_fluo_reconstruction(oversamplings, widthsToTest, nbAngle, outgoingraypointsmethods, nbIter): """ benchmark the fluo reconstruction :param oversamplings: the list of oversamplings to test :param widthsToTest: the list of phantom width to test :param nbAngle: the number of projection to test :param outgoingraypointsmethods: the list of outgoingrayalgorithm to test :param nbIter: the number of iteration to test (one iteration launch nbAngle back projection) """ times = {} for oversamp in oversamplings : for obmc in outgoingraypointsmethods: bid = "method="+ str(obmc) + "_oversamp=" + str(oversamp) times[bid] = {} for width in widthsToTest : sinogram, angles, absMat, selfAbsMat, time = exampleutils.produce_fluo_SheppLogan_sinogram(oversamp, nbAngle, detSetup, width, save=False) times[bid][width] = exampleutils.make_fluo_reconstruction(sinogram, angles, absMat, selfAbsMat, detSetup, nbIter, oversamp)[1] print("fluo reconstruction - execution time for " + str(oversamp) + " oversampling, with = " + str(width) + ", method = " + str(obmc) + " toke : " + str(times[bid][width])) if savingToPdf : for obmc in outgoingraypointsmethods: title=("fluo_benchmark_for_reconstruction_method_" + str(obmc) + "_on_" + str(nbAngle) + "_angles_and_" + str(nbIter) + "_iteration") plot = plot2D(title=title ) for curve in times: plot.addCurve(widthsToTest, times[curve].values(), legend=curve ) plot.legendsDockWidget.show() plot.saveGraph(filename=(title + ".png"), fileFormat="png") plot.hide()
if __name__ == "__main__": if ('--saving' or '--s') in sys.argv : savingToPdf=True # benchmark the transmission oversamplings = [1, 2, 4, 8] widths = [16, 32, 64, 128, 356, 1024] methodsBeamsGeneral = [raypointsmethod.withInterpolation ] methodsOutgoing = [ outgoingrayalgorithm.rawApproximation, outgoingrayalgorithm.createOneRayPerSamplePoint, outgoingrayalgorithm.matriceSubdivision] if savingToPdf : global app # QApplication must be global to avoid seg fault on quit app = qt.QApplication([]) # TODO : compute mean error benchmark_tx_projection(oversamplings, widths, nbAngle = 100) benchmark_tx_reconstruction(oversamplings, widths, nbAngle = 100, nbIter=1) detPos = (1000, 0, 0) detectorRadius = 10 detSetup = [(detPos, detectorRadius)] benchmark_fluo_projection(oversamplings, widths, 100, methodsOutgoing) benchmark_fluo_reconstruction(oversamplings, widths, 100, methodsOutgoing, nbIter=1) if savingToPdf : app.exec_()