コード例 #1
0
ファイル: decode2.py プロジェクト: migurski/paperwalking
def _blob_matches_secondary(blobs, aed_match):
    """ Generate known matches for AED (bottom) and paper-specific triangle groups.
    """
    from dimensions import feature_g_landscape_ltr, point_G_landscape_ltr, feature_f_landscape_ltr, point_F_landscape_ltr
    from dimensions import feature_g_portrait_ltr, point_G_portrait_ltr, feature_f_portrait_ltr, point_F_portrait_ltr
    from dimensions import feature_g_landscape_a4, point_G_landscape_a4, feature_f_landscape_a4, point_F_landscape_a4
    from dimensions import feature_g_portrait_a4, point_G_portrait_a4, feature_f_portrait_a4, point_F_portrait_a4
    from dimensions import feature_g_landscape_a3, point_G_landscape_a3, feature_f_landscape_a3, point_F_landscape_a3
    from dimensions import feature_g_portrait_a3, point_G_portrait_a3, feature_f_portrait_a3, point_F_portrait_a3

    from dimensions import min_size, theta_tol, ratio_tol

    features_fg = (
        (feature_g_landscape_ltr, point_G_landscape_ltr, feature_f_landscape_ltr, point_F_landscape_ltr),
        (feature_g_portrait_ltr,  point_G_portrait_ltr,  feature_f_portrait_ltr,  point_F_portrait_ltr),
        (feature_g_landscape_a4,  point_G_landscape_a4,  feature_f_landscape_a4,  point_F_landscape_a4),
        (feature_g_portrait_a4,   point_G_portrait_a4,   feature_f_portrait_a4,   point_F_portrait_a4),
        (feature_g_landscape_a3,  point_G_landscape_a3,  feature_f_landscape_a3,  point_F_landscape_a3),
        (feature_g_portrait_a3,   point_G_portrait_a3,   feature_f_portrait_a3,   point_F_portrait_a3)
      )

    for (feature_g, point_G, feature_f, point_F) in features_fg:
        g_theta, g_ratio = feature_g.theta, feature_g.ratio
        g_bounds = theta_ratio_bounds(g_theta, theta_tol, g_ratio, ratio_tol)
        g_matches = blobs2feats_fitted(aed_match.s1, aed_match.s2, blobs, *g_bounds)
        
        for g_tuple in g_matches:
            i0, j0, k0 = g_tuple[0:3]
            g_match = MatchedFeature(feature_g, blobs[i0], blobs[j0], blobs[k0])
            
            if not g_match.fits(aed_match):
                continue
            
            aed_blobs = (aed_match.s1, aed_match.s2)
            
            if g_match.s1 in aed_blobs and g_match.s2 in aed_blobs:
                blob_G = g_match.s3
            elif g_match.s1 in aed_blobs and g_match.s3 in aed_blobs:
                blob_G = g_match.s2
            elif g_match.s2 in aed_blobs and g_match.s3 in aed_blobs:
                blob_G = g_match.s1
            else:
                raise Exception('what?')
            
            #print >> stderr, '    Found a match for point G -', (i0, j0, k0)

            #
            # We think we have a match for point G, now check for point F.
            #
            
            f_theta, f_ratio = feature_f.theta, feature_f.ratio
            f_bounds = theta_ratio_bounds(f_theta, theta_tol, f_ratio, ratio_tol)
            f_matches = blobs2feats_fitted(blob_G, aed_match.s2, blobs, *f_bounds)
            
            for f_tuple in f_matches:
                i1, j1, k1 = f_tuple[0:3]
                f_match = MatchedFeature(feature_f, blobs[i1], blobs[j1], blobs[k1])
                
                if not f_match.fits(g_match):
                    continue

                #print >> stderr, '     Found a match for point F -', (i1, j1, k1), point_F
                
                #
                # Based on the identity of point_F, we can find paper size and orientation.
                #
                yield f_match, point_F
コード例 #2
0
def _blob_matches_secondary(blobs, aed_match):
    """ Generate known matches for AED (bottom) and paper-specific triangle groups.
    """
    from dimensions import feature_g_landscape_ltr, point_G_landscape_ltr, feature_f_landscape_ltr, point_F_landscape_ltr
    from dimensions import feature_g_portrait_ltr, point_G_portrait_ltr, feature_f_portrait_ltr, point_F_portrait_ltr
    from dimensions import feature_g_landscape_a4, point_G_landscape_a4, feature_f_landscape_a4, point_F_landscape_a4
    from dimensions import feature_g_portrait_a4, point_G_portrait_a4, feature_f_portrait_a4, point_F_portrait_a4
    from dimensions import feature_g_landscape_a3, point_G_landscape_a3, feature_f_landscape_a3, point_F_landscape_a3
    from dimensions import feature_g_portrait_a3, point_G_portrait_a3, feature_f_portrait_a3, point_F_portrait_a3

    from dimensions import min_size, theta_tol, ratio_tol

    features_fg = ((feature_g_landscape_ltr, point_G_landscape_ltr,
                    feature_f_landscape_ltr, point_F_landscape_ltr),
                   (feature_g_portrait_ltr, point_G_portrait_ltr,
                    feature_f_portrait_ltr, point_F_portrait_ltr),
                   (feature_g_landscape_a4, point_G_landscape_a4,
                    feature_f_landscape_a4, point_F_landscape_a4),
                   (feature_g_portrait_a4, point_G_portrait_a4,
                    feature_f_portrait_a4, point_F_portrait_a4),
                   (feature_g_landscape_a3, point_G_landscape_a3,
                    feature_f_landscape_a3, point_F_landscape_a3),
                   (feature_g_portrait_a3, point_G_portrait_a3,
                    feature_f_portrait_a3, point_F_portrait_a3))

    for (feature_g, point_G, feature_f, point_F) in features_fg:
        g_theta, g_ratio = feature_g.theta, feature_g.ratio
        g_bounds = theta_ratio_bounds(g_theta, theta_tol, g_ratio, ratio_tol)
        g_matches = blobs2feats_fitted(aed_match.s1, aed_match.s2, blobs,
                                       *g_bounds)

        for g_tuple in g_matches:
            i0, j0, k0 = g_tuple[0:3]
            g_match = MatchedFeature(feature_g, blobs[i0], blobs[j0],
                                     blobs[k0])

            if not g_match.fits(aed_match):
                continue

            aed_blobs = (aed_match.s1, aed_match.s2)

            if g_match.s1 in aed_blobs and g_match.s2 in aed_blobs:
                blob_G = g_match.s3
            elif g_match.s1 in aed_blobs and g_match.s3 in aed_blobs:
                blob_G = g_match.s2
            elif g_match.s2 in aed_blobs and g_match.s3 in aed_blobs:
                blob_G = g_match.s1
            else:
                raise Exception('what?')

            #print '    Found a match for point G -', (i0, j0, k0)

            #
            # We think we have a match for point G, now check for point F.
            #

            f_theta, f_ratio = feature_f.theta, feature_f.ratio
            f_bounds = theta_ratio_bounds(f_theta, theta_tol, f_ratio,
                                          ratio_tol)
            f_matches = blobs2feats_fitted(blob_G, aed_match.s2, blobs,
                                           *f_bounds)

            for f_tuple in f_matches:
                i1, j1, k1 = f_tuple[0:3]
                f_match = MatchedFeature(feature_f, blobs[i1], blobs[j1],
                                         blobs[k1])

                if not f_match.fits(g_match):
                    continue

                #print '     Found a match for point F -', (i1, j1, k1), point_F

                #
                # Based on the identity of point_F, we can find paper size and orientation.
                #
                yield f_match, point_F
コード例 #3
0
ファイル: decode2.py プロジェクト: migurski/paperwalking
def _blob_matches_primary(blobs):
    """ Generate known matches for DBC (top) and AED (bottom) triangle pairs.
    """
    from dimensions import feature_dbc, feature_dab, feature_aed, feature_eac
    from dimensions import min_size, theta_tol, ratio_tol

    dbc_theta, dbc_ratio = feature_dbc.theta, feature_dbc.ratio
    dab_theta, dab_ratio = feature_dab.theta, feature_dab.ratio
    aed_theta, aed_ratio = feature_aed.theta, feature_aed.ratio
    eac_theta, eac_ratio = feature_eac.theta, feature_eac.ratio
    
    dbc_matches = blobs2features(blobs, min_size, *theta_ratio_bounds(dbc_theta, theta_tol, dbc_ratio, ratio_tol))
    
    seen_groups, max_skipped, skipped_groups = set(), 100, 0
    
    for dbc_tuple in dbc_matches:
        i0, j0, k0 = dbc_tuple[0:3]
        dbc_match = MatchedFeature(feature_dbc, blobs[i0], blobs[j0], blobs[k0])
    
        #print >> stderr, 'Found a match for DBC -', (i0, j0, k0)
        
        dab_matches = blobs2feats_limited([blobs[i0]], blobs, [blobs[j0]], *theta_ratio_bounds(dab_theta, theta_tol, dab_ratio, ratio_tol))
        
        for dab_tuple in dab_matches:
            i1, j1, k1 = i0, dab_tuple[1], j0
            dab_match = MatchedFeature(feature_dab, blobs[i1], blobs[j1], blobs[k1])
            
            if not dab_match.fits(dbc_match):
                continue
            
            #print >> stderr, ' Found a match for DAB -', (i1, j1, k1)
            
            #
            # We think we have a match for points A-D, now check for point E.
            #
            
            aed_matches = blobs2feats_limited([blobs[j1]], blobs, [blobs[i1]], *theta_ratio_bounds(aed_theta, theta_tol, aed_ratio, ratio_tol))
            
            for aed_tuple in aed_matches:
                i2, j2, k2 = j1, aed_tuple[1], i1
                aed_match = MatchedFeature(feature_aed, blobs[i2], blobs[j2], blobs[k2])
                
                if not aed_match.fits(dbc_match):
                    continue
                
                if not aed_match.fits(dab_match):
                    continue
                
                #print >> stderr, '  Found a match for AED -', (i2, j2, k2)
                
                #
                # We now know we have a three-triangle match; try a fourth to verify.
                # Use the very small set of blobs from the current set of matches.
                #
                
                _blobs = [blobs[n] for n in set((i0, j0, k0) + (i1, j1, k1) + (i2, j2, k2))]
                eac_matches = blobs2features(_blobs, min_size, *theta_ratio_bounds(eac_theta, theta_tol, eac_ratio, ratio_tol))
                
                for eac_tuple in eac_matches:
                    i3, j3, k3 = eac_tuple[0:3]
                    eac_match = MatchedFeature(feature_eac, _blobs[i3], _blobs[j3], _blobs[k3])
                    
                    if not eac_match.fits(dbc_match):
                        continue
                    
                    if not eac_match.fits(dab_match):
                        continue
                    
                    if not eac_match.fits(aed_match):
                        continue
                    
                    #print >> stderr, '   Confirmed match with EAC -', (i3, j3, k3)
                    
                    yield dbc_match, aed_match
コード例 #4
0
def _blob_matches_primary(blobs):
    """ Generate known matches for DBC (top) and AED (bottom) triangle pairs.
    """
    from dimensions import feature_dbc, feature_dab, feature_aed, feature_eac
    from dimensions import min_size, theta_tol, ratio_tol

    dbc_theta, dbc_ratio = feature_dbc.theta, feature_dbc.ratio
    dab_theta, dab_ratio = feature_dab.theta, feature_dab.ratio
    aed_theta, aed_ratio = feature_aed.theta, feature_aed.ratio
    eac_theta, eac_ratio = feature_eac.theta, feature_eac.ratio

    dbc_matches = blobs2features(
        blobs, min_size,
        *theta_ratio_bounds(dbc_theta, theta_tol, dbc_ratio, ratio_tol))

    seen_groups, max_skipped, skipped_groups = set(), 100, 0

    for dbc_tuple in dbc_matches:
        i0, j0, k0 = dbc_tuple[0:3]
        dbc_match = MatchedFeature(feature_dbc, blobs[i0], blobs[j0],
                                   blobs[k0])

        #print 'Found a match for DBC -', (i0, j0, k0)

        dab_matches = blobs2feats_limited([blobs[i0]], blobs, [blobs[j0]],
                                          *theta_ratio_bounds(
                                              dab_theta, theta_tol, dab_ratio,
                                              ratio_tol))

        for dab_tuple in dab_matches:
            i1, j1, k1 = i0, dab_tuple[1], j0
            dab_match = MatchedFeature(feature_dab, blobs[i1], blobs[j1],
                                       blobs[k1])

            if not dab_match.fits(dbc_match):
                continue

            #print ' Found a match for DAB -', (i1, j1, k1)

            #
            # We think we have a match for points A-D, now check for point E.
            #

            aed_matches = blobs2feats_limited([blobs[j1]], blobs, [blobs[i1]],
                                              *theta_ratio_bounds(
                                                  aed_theta, theta_tol,
                                                  aed_ratio, ratio_tol))

            for aed_tuple in aed_matches:
                i2, j2, k2 = j1, aed_tuple[1], i1
                aed_match = MatchedFeature(feature_aed, blobs[i2], blobs[j2],
                                           blobs[k2])

                if not aed_match.fits(dbc_match):
                    continue

                if not aed_match.fits(dab_match):
                    continue

                #print '  Found a match for AED -', (i2, j2, k2)

                #
                # We now know we have a three-triangle match; try a fourth to verify.
                # Use the very small set of blobs from the current set of matches.
                #

                _blobs = [
                    blobs[n]
                    for n in set((i0, j0, k0) + (i1, j1, k1) + (i2, j2, k2))
                ]
                eac_matches = blobs2features(
                    _blobs, min_size,
                    *theta_ratio_bounds(eac_theta, theta_tol, eac_ratio,
                                        ratio_tol))

                for eac_tuple in eac_matches:
                    i3, j3, k3 = eac_tuple[0:3]
                    eac_match = MatchedFeature(feature_eac, _blobs[i3],
                                               _blobs[j3], _blobs[k3])

                    if not eac_match.fits(dbc_match):
                        continue

                    if not eac_match.fits(dab_match):
                        continue

                    if not eac_match.fits(aed_match):
                        continue

                    #print '   Confirmed match with EAC -', (i3, j3, k3)

                    yield dbc_match, aed_match