Skip to content
Snippets Groups Projects
gtReconstructTestCase.m 8.62 KiB
Newer Older
function [result_viewer, result, post_result, algo] = gtReconstructTestCase(test_data, det_index, varargin)

    if (~exist('det_index', 'var') || isempty(det_index))
        det_index = 1;
    end

    conf = struct( ...
        'do_plot', false, ...
        'single_orientation', false, ...
        'rspace_super_resolution', 1);
    conf = parse_pv_pairs(conf, varargin);
    rec_opts = gtReconstruct6DGetParamenters(test_data.parameters);

    if (~isfield(test_data, 'gv'))
        test_data.gv.dm = gtDefDmvol2Gvdm(test_data.dmvol);
    end
    gvdm = test_data.gv.dm(:, test_data.gv.used_ind);
Nicola Vigano's avatar
Nicola Vigano committed
%     gvcs = test_data.gv.cs(:, test_data.gv.used_ind);
%     if (~isfield(test_data, 'intvol'))
%         size_int_vol = [ ...
%             size(test_data.dmvol, 1), ...
%             size(test_data.dmvol, 2), ...
%             size(test_data.dmvol, 3) ];
%         test_data.intvol = ones(size_int_vol, 'like', test_data.dmvol);
%     end
%     gvpow = test_data.intvol(test_data.gv.used_ind);
    cryst = test_data.parameters.cryst(test_data.grain.phaseid);
    symm = gtCrystGetSymmetryOperators(cryst.crystal_system, cryst.spacegroup);
    gtGetMaxDisorientation(gvdm, symm, 'diameter_average');
    gtGetMaxDisorientation(gvdm, symm, 'diameter_zero');
Nicola Vigano's avatar
Nicola Vigano committed
        f = figure();
        ax = axes('parent', f);
        hold(ax, 'on')
        scatter3(ax, gvdm(1, 1:10:end)', gvdm(2, 1:10:end)', gvdm(3, 1:10:end)', 20);
    end

    if (conf.rspace_super_resolution > 1)
        test_data.parameters.recgeo(det_index(1)).voxsize = ...
            test_data.parameters.recgeo(det_index(1)).voxsize / conf.rspace_super_resolution;

        test_data.grain.proj(det_index(1)).vol_size_x = ...
            test_data.grain.proj(det_index(1)).vol_size_x * conf.rspace_super_resolution;
        test_data.grain.proj(det_index(1)).vol_size_y = ...
            test_data.grain.proj(det_index(1)).vol_size_y * conf.rspace_super_resolution;
        test_data.grain.proj(det_index(1)).vol_size_z = ...
            test_data.grain.proj(det_index(1)).vol_size_z * conf.rspace_super_resolution;
    end

    sampler = GtOrientationSampling(test_data.parameters, test_data.grain, 'detector_index', det_index(1));
Nicola Vigano's avatar
Nicola Vigano committed
        % Testing ability of GtOrientationSampling to detect the correct
        % orientation-space boundingbox
        sampler.make_grid_resolution(rec_opts.ospace_resolution, rec_opts.max_grid_edge_points);
        scatter3(ax, sampler.R_vectors(:, 1), sampler.R_vectors(:, 2), sampler.R_vectors(:, 3), 20, 'r');
Nicola Vigano's avatar
Nicola Vigano committed
        drawnow();

        sampler = GtOrientationSampling(test_data.parameters, test_data.grain, 'detector_index', det_index(1));
Nicola Vigano's avatar
Nicola Vigano committed
    end

    is_topotomo = (numel(det_index) == 1) && strcmpi(test_data.parameters.acq(det_index).type, 'topotomo');

    gvdm_lims = [min(gvdm, [], 2), max(gvdm, [], 2)];
    gvdm_lims(:, 1) = gvdm_lims(:, 1) - tand(rec_opts.ospace_resolution / 4);
    gvdm_lims(:, 2) = gvdm_lims(:, 2) + tand(rec_opts.ospace_resolution / 4);
        gvdm_lims(3, :) = [-1, 1] * tand(rec_opts.ospace_resolution / 4);
        rec_opts.max_grid_edge_points = [rec_opts.max_grid_edge_points([1 1]), 1];

        error_axes = 1:2;
    else
        error_axes = 1:3;
    for ii_d = 1:numel(det_index)
        % Using the optimal orientation-space boundingbox
            sampler.make_undeformed_sampling('det_ind', det_index(ii_d));
        else
            sampler.make_grid_resolution(rec_opts.ospace_resolution, ...
                rec_opts.max_grid_edge_points, gvdm_lims, ...
                'oversize', 1, 'det_ind', det_index(ii_d));
        end

        if (rec_opts.ospace_super_sampling > 1)
            sampler.make_supersampling_simple_grid([1 2 3], rec_opts.ospace_super_sampling, det_index(ii_d));
        end
    end
Nicola Vigano's avatar
Nicola Vigano committed
%         ospace_bb = [min(gvdm, [], 2) - 0.002, max(gvdm, [], 2) + 0.002];
%         sampler.make_grid('cubic', 3, ospace_bb, rec_opts.ospace_oversize);
        scatter3(ax, sampler.R_vectors(:, 1), sampler.R_vectors(:, 2), sampler.R_vectors(:, 3), 20, 'g');
Nicola Vigano's avatar
Nicola Vigano committed
        hold(ax, 'off')
        drawnow();
    end
    if (conf.single_orientation)
        detgeo = test_data.parameters.detgeo(det_index(1));
        recgeo = test_data.parameters.recgeo(det_index(1));
        samgeo = test_data.parameters.samgeo;
        labgeo = test_data.parameters.labgeo;

        proj = test_data.grain.proj(det_index(1));
        ab = test_data.grain.allblobs(det_index(1));
        proj.num_iter = rec_opts.num_iter;
        proj.stack = proj.stack(:, proj.selected, :);

        proj.num_cols = size(proj.stack, 1);
        proj.num_rows = size(proj.stack, 3);

        ab_sel = proj.ondet(proj.included(proj.selected));
        dvecsam = ab.dvecsam(ab_sel, :);
        rot_w_l2s = ab.srot(:, :, ab_sel);
        bbuim = cat(1, proj.bl(proj.selected).bbuim);
        bbvim = cat(1, proj.bl(proj.selected).bbvim);
        bb_pos = [bbuim(:, 1), bbvim(:, 1), ...
            bbuim(:, 2) - bbuim(:, 1) + 1, ...
            bbvim(:, 2) - bbvim(:, 1) + 1, ];

        volume_size = [proj.vol_size_y, proj.vol_size_x, proj.vol_size_z];
        spacing = mean(volume_size) * (rec_opts.rspace_oversize - 1);
        volume_size = ceil(volume_size + spacing);
        proj.vol_size_y = volume_size(1);
        proj.vol_size_x = volume_size(2);
        proj.vol_size_z = volume_size(3);

        proj.geom = gtGeoProjForReconstruction(dvecsam, ...
            rot_w_l2s, test_data.grain.center, bb_pos, [], ...
            detgeo, labgeo, samgeo, recgeo, 'ASTRA_grain');

        vols = { gtAstra3D(proj, 'is_cone', false) };
    else
        algo = gtReconstruct6DLaunchAlgorithm(sampler, rec_opts, ...
            test_data.parameters, 'det_index', det_index);

        vols = algo.getCurrentSolution();
    end
    or_sizes = sampler.get_orientation_sampling_size();

    [avg_R_vecs, avg_R_vecs_int, stddev_R_vecs] = sampler.getAverageOrientations(vols);
    avg_R_vec = sampler.getAverageOrientation(vols);
    s_g_odf = reshape(sampler.getODF(vols), or_sizes);

    vol_size = size(avg_R_vecs_int);
    shift = gtFwdSimComputeVolumeShifts(test_data.grain.proj, test_data.parameters, vol_size, det_index(1));

    [kam, gam] = gtDefComputeKernelAverageMisorientation(avg_R_vecs, avg_R_vecs_int);
    [igm, gos] = gtDefComputeIntraGranularMisorientation(avg_R_vecs, avg_R_vecs_int, 'R_vector', test_data.grain.R_vector);

    % Restoring initial volume size (depending on the rounding)
    if (rec_opts.volume_downscaling > 1)
        fprintf('Expanding volumes..')
        c = tic();
        avg_R_vecs_int = gtMathsUpsampleVolume(avg_R_vecs_int, rec_opts.volume_downscaling);
        avg_R_vecs = gtMathsUpsampleVolume(avg_R_vecs, rec_opts.volume_downscaling);
        stddev_R_vecs = gtMathsUpsampleVolume(stddev_R_vecs, rec_opts.volume_downscaling);
        kam = gtMathsUpsampleVolume(kam, rec_opts.volume_downscaling);
        igm = gtMathsUpsampleVolume(igm, rec_opts.volume_downscaling);

        vols = gtMathsUpsampleVolume(vols, rec_opts.volume_downscaling);
        fprintf('\b\b: Done (%f seconds).\n', toc(c))
    end

    result = gt6DAssembleResult(test_data, sampler.get_orientations(), vols, det_index);
    ODF6D = gtReconstructDataStructureDefinition('VOL6D');
    ODF6D.options = rec_opts;
    if (~conf.single_orientation)
        ODF6D.compute_statistics = algo.get_statistics();
    end

    ODF6D.voxels_avg_R_vectors = avg_R_vecs;
    ODF6D.intensity = avg_R_vecs_int;
    ODF6D.shift = shift;
    ODF6D.R_vectors = sampler.get_R_vectors();
    ODF6D.voxels_stddev_R_vectors = stddev_R_vecs;
    ODF6D.single_grain_ODF = s_g_odf;
    ODF6D.single_grain_avg_R_vector = avg_R_vec;
    ODF6D.kernel_average_misorientation = kam;
    ODF6D.intra_granular_misorientation = igm;

    result.ODF6D = ODF6D;
    t = GtThreshold(test_data.parameters);
    t.param.rec.thresholding.do_region_prop = false;
    t.param.rec.thresholding.use_levelsets = true;
    gr_rec = struct('vol', avg_R_vecs_int, 'shift', shift);
    result.SEG = t.thresholdAutoSingleGrain(gr_rec);

    post_result = gt6DPostProcessOrientationSpread(test_data, result, true, error_axes);
    if (~conf.single_orientation)
        [proj_blobs, proj_spots] = algo.getProjectionOfCurrentSolution();
        post_result.ODF6D = struct( ...
            'orientation_volumes', {vols}, ...
            'fwd_projected_blobs', {proj_blobs}, ...
            'fwd_projected_spots', {proj_spots}, ...
            'grain_average_misorientation', {gam}, ...
            'grain_orientation_spread', {gos}, ...
            'compute_statistics', {algo.get_statistics()} );
    end
    result_viewer = GtOrientationResultView(test_data, result, post_result);