J/MNRAS/463/2939 Automatic galaxy detection & classification (Tramacere+, 2016)
ASTErIsM: application of topometric clustering algorithms in automatic galaxy
detection and classification.
Tramacere A., Paraficz D., Dubath P., Kneib J.-P., Courbin F.
<Mon. Not. R. Astron. Soc., 463, 2939-2957 (2016)>
=2016MNRAS.463.2939T 2016MNRAS.463.2939T (SIMBAD/NED BibCode)
ADC_Keywords: Galaxy catalogs ; Morphology
Keywords: methods: data analysis - methods: statistical - catalogues -
galaxies: elliptical and lenticular, cD - galaxies: general -
galaxies: spiral
Abstract:
We present a study on galaxy detection and shape classification using
topometric clustering algorithms. We first use the DBSCAN algorithm to
extract, from CCD frames, groups of adjacent pixels with significant
fluxes and we then apply the DENCLUE algorithm to separate the
contributions of overlapping sources. The DENCLUE separation is based
on the localization of pattern of local maxima, through an iterative
algorithm, which associates each pixel to the closest local maximum.
Our main classification goal is to take apart elliptical from spiral
galaxies. We introduce new sets of features derived from the
computation of geometrical invariant moments of the pixel group shape
and from the statistics of the spatial distribution of the DENCLUE
local maxima patterns. Ellipticals are characterized by a single group
of local maxima, related to the galaxy core, while spiral galaxies
have additional groups related to segments of spiral arms. We use two
different supervised ensemble classification algorithms: Random Forest
and Gradient Boosting. Using a sample of ~=24000 galaxies taken
from the Galaxy Zoo 2 main sample with spectroscopic redshifts, and we
test our classification against the Galaxy Zoo 2 catalogue. We find
that features extracted from our pipeline give, on average, an
accuracy of ~=93 per cent, when testing on a test set with a
size of 20 per cent of our full data set, with features deriving from
the angular distribution of density attractor ranking at the top of
the discrimination power.
Description:
The results presented in this work show the successful application of
ASTErIsM software, based on topometric clustering algorithms (dbscan
and denclue), to automatic galaxy detection and shape classification.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
catalog.dat 1156 24635 Automatic galaxy detection and shape
classification catalog
--------------------------------------------------------------------------------
Byte-by-byte Description of file: catalog.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 20 I20 --- specobjid Match to the DR8 spectrum object (specobjid)
22- 30 F9.5 deg RAdeg Right ascension (J2000.0) (RAdeg)
32- 41 F10.6 deg DEdeg Declination (J2000.0]) (DEdeg)
43 A1 --- GZ2class GZ class label for the E/S classification
(gz2class) (1)
45- 46 A2 --- t01 Label for task 01 of the GZ2 decision
tree (t01) (2)
48- 49 A2 --- t04 Label for task 04 of the GZ2 decision
tree (t04) (3)
51- 52 I2 --- idCluster [-1/0] ID of the ASTErIsM-detected cluster
corresponding to the GZ2 source
(idCluster) (4)
54- 58 I5 --- gpsizeic ?=-99 Geometrical features for the initial
cluster, number of pixels of the cluster
(geompixsize_ic)
60- 69 F10.6 --- geccic ?=-99 Geometrical features for the initial
cluster, eccentricity of the cluster
containment ellipsoid (geomeccic)
71- 79 F9.4 --- gcompic ?=-99 Geometrical features for the initial
cluster, geometrical compactness
(geomcompic)
81- 89 F9.5 --- garic ?=-99 Geometrical features for the initial
cluster, aspect ratio of the minimal
rectangular box enclosing the source
cluster (geomaric)
91-100 F10.6 --- gcontric ?=-99 Geometrical features for the initial
cluster, contour ratio
(geomcontourratio_ic)
102-111 F10.6 --- gratioic ?=-99 Geometrical features for the initial
cluster, area ratio (geomratioic)
113-120 F8.4 --- grmaxic ?=-99 Geometrical features for the initial
cluster, rmax of the cluster (geomrmax_ic)
122-126 I5 --- gpsizedepr ?=-99 Geometrical features for the deprojected
cluster, number of pixels of the cluster
(geompixsize_depr)
128-137 F10.6 --- geccdepr ?=-99 Geometrical features for the deprojected
cluster, eccentricity of the cluster
containment ellipsoid (geomeccdepr)
139-147 F9.4 --- gcompdepr ?=-99 Geometrical features for the deprojected
cluster, geometrical compactness
(geomcompdepr)
149-157 F9.5 --- gardepr ?=-99 Geometrical features for the deprojected
cluster, aspect ratio of the minimal
rectangular box enclosing the source cluster
(geomardepr)
159-168 F10.6 --- gcontrdepr ?=-99 Geometrical features for the deprojected
cluster, contour ratio
(geomcontourratio_depr)
170-179 F10.6 --- gratiodepr ?=-99 Geometrical features for the deprojected
cluster, area ratio (geomratiodepr)
181-189 F9.5 --- grmaxdepr ?=-99 Geometrical features for the deprojected
cluster, rmax of the cluster
(geomrmax_depr)
191-195 I5 --- gpsizeunsh ?=-99 Geometrical features for the unsharp
cluster, number of pixels of the cluster
(geompixsize_unsh)
197-206 F10.6 --- geccunsh ?=-99 Geometrical features for the unsharp
cluster, eccentricity of the cluster
containment ellipsoid (geomeccunsh)
208-216 F9.4 --- gcompunsh ?=-99 Geometrical features for the unsharp
cluster, geometrical compactness
(geomcompunsh)
218-226 F9.5 --- garunsh ?=-99 Geometrical features for the unsharp
cluster, aspect ratio of the minimal
rectangular box enclosing the source cluster
(geomarunsh)
228-237 F10.6 --- gcontrunsh ?=-99 Geometrical features for the unsharp
cluster, contour ratio
(geomcontourratio_unsh)
239-248 F10.6 --- gratiounsh ?=-99 Geometrical features for the unsharp
cluster, area ratio (geomratiounsh)
250-259 F10.6 --- grmaxunsh ?=-99 Geometrical features for the unsharp
cluster, rmax of the cluster
(geomrmax_unsh)
261-270 F10.6 --- clHu0ic ?=-99 logarithm of Hu moment 0 of the initial
cluster contour (cntlogHu0ic)
272-281 F10.6 --- clHu1ic ?=-99 logarithm of Hu moment 1 of the initial
cluster contour (cntlogHu1ic)
283-291 F9.5 --- clHu2ic ?=-99 logarithm of Hu moment 2 of the initial
cluster contour (cntlogHu2ic)
293-301 F9.5 --- clHu3ic ?=-99 logarithm of Hu moment 3 of the initial
cluster contour (cntlogHu3ic)
303-311 F9.5 --- clHu4ic ?=-99 logarithm of Hu moment 4 of the initial
cluster contour (cntlogHu4ic)
313-321 F9.5 --- clHu5ic ?=-99 logarithm of Hu moment 5 of the initial
cluster contour (cntlogHu5ic)
323-331 F9.5 --- clHu6ic ?=-99 logarithm of Hu moment 6 of the initial
cluster contour (cntlogHu6ic)
333-342 F10.6 --- ilHu0ic ?=-99 logarithm of Hu moment 0 of the initial
cluster contour (imglogHu0ic)
344-353 F10.6 --- ilHu1ic ?=-99 logarithm of Hu moment 1 of the initial
cluster image (imglogHu1ic)
355-364 F10.6 --- ilHu2ic ?=-99 logarithm of Hu moment 2 of the initial
cluster image (imglogHu2ic)
366-375 F10.6 --- ilHu3ic ?=-99 logarithm of Hu moment 3 of the initial
cluster image (imglogHu3ic)
377-386 F10.6 --- ilHu4ic ?=-99 logarithm of Hu moment 4 of the initial
cluster image (imglogHu4ic)
388-397 F10.6 --- ilHu5ic ?=-99 logarithm of Hu moment 5 of the initial
cluster image (imglogHu5ic)
399-408 F10.6 --- ilHu6ic ?=-99 logarithm of Hu moment 6 of the initial
cluster image (imglogHu6ic)
410-419 F10.6 --- aplHu0ic ?=-99 logarithm of Hu moment 0 for the density
attractors in polar coordinates
(attrpolarlogHu0_ic)
421-430 F10.6 --- aplHu1ic ?=-99 logarithm of Hu moment 1 for the density
attractors in polar coordinates
(attrpolarlogHu1_ic)
432-441 F10.6 --- aplHu2ic ?=-99 logarithm of Hu moment 2 for the density
attractors in polar coordinates
(attrpolarlogHu2_ic)
443-452 F10.6 --- aplHu3ic ?=-99 logarithm of Hu moment 3 for the density
attractors in polar coordinates
(attrpolarlogHu3_ic)
454-463 F10.6 --- aplHu4ic ?=-99 logarithm of Hu moment 4 for the density
attractors in polar coordinates
(attrpolarlogHu4_ic)
465-474 F10.6 --- aplHu5ic ?=-99 logarithm of Hu moment 5 for the density
attractors in polar coordinates
(attrpolarlogHu5_ic)
476-485 F10.6 --- aplHu6ic ?=-99 logarithm of Hu moment 6 for the density
attractors in polar coordinates
(attrpolarlogHu6_ic)
487-496 F10.6 --- alHu0ic ?=-99 logarithm of Hu moment 0 for the density
attractors in Cartesian coordinates
(attrlogHu0ic)
498-507 F10.6 --- alHu1ic ?=-99 logarithm of Hu moment 1 for the density
attractors in Cartesian coordinates
(attrlogHu1ic)
509-518 F10.6 --- alHu2ic ?=-99 logarithm of Hu moment 2 for the density
attractors in Cartesian coordinates
(attrlogHu2ic)
520-529 F10.6 --- alHu3ic ?=-99 logarithm of Hu moment 3 for the density
attractors in Cartesian coordinates
(attrlogHu3ic)
531-540 F10.6 --- alHu4ic ?=-99 logarithm of Hu moment 4 for the density
attractors in Cartesian coordinates
(attrlogHu4ic)
542-551 F10.6 --- alHu5ic ?=-99 logarithm of Hu moment 5 for the density
attractors in Cartesian coordinates
(attrlogHu5ic)
553-562 F10.6 --- alHu6ic ?=-99 logarithm of Hu moment 6 for the density
attractors in Cartesian coordinates
(attrlogHu6ic)
564-573 F10.6 --- clHu0depr ?=-99 logarithm of Hu moment 0 for the
deprojected cluster contour
(cntlogHu0depr)
575-584 F10.6 --- clHu1depr ?=-99 logarithm of Hu moment 1 for the
deprojected cluster contour
(cntlogHu1depr)
586-594 F9.5 --- clHu2depr ?=-99 logarithm of Hu moment 2 for the
deprojected cluster contour
(cntlogHu2depr)
596-604 F9.5 --- clHu3depr ?=-99 logarithm of Hu moment 3 for the
deprojected cluster contour
(cntlogHu3depr)
606-614 F9.5 --- clHu4depr ?=-99 logarithm of Hu moment 4 for the
deprojected cluster contour
(cntlogHu4depr)
616-624 F9.5 --- clHu5depr ?=-99 logarithm of Hu moment 5 for the
deprojected cluster contour
(cntlogHu5depr)
626-634 F9.5 --- clHu6depr ?=-99 logarithm of Hu moment 6 for the
deprojected cluster contour
(cntlogHu6depr)
636-645 F10.6 --- ilHu0depr ?=-99 logarithm of Hu moment 0 for the
deprojected cluster image (imglogHu0depr)
647-656 F10.6 --- ilHu1depr ?=-99 logarithm of Hu moment 1 for the
deprojected cluster image (imglogHu1depr)
658-666 F9.5 --- ilHu2depr ?=-99 logarithm of Hu moment 2 for the
deprojected cluster image (imglogHu2depr)
668-676 F9.5 --- ilHu3epr ?=-99 logarithm of Hu moment 3 for the
deprojected cluster image (imglogHu3depr)
678-686 F9.5 --- ilHu4depr ?=-99 logarithm of Hu moment 4 for the
deprojected cluster image (imglogHu4depr)
688-696 F9.5 --- ilHu5depr ?=-99 logarithm of Hu moment 5 for the
deprojected cluster image (imglogHu5depr)
698-706 F9.5 --- ilHu6depr ?=-99 logarithm of Hu moment 6 for the
deprojected cluster image (imglogHu6depr)
708-717 F10.6 --- Giniic ?=-99 Morphological features for the initial
cluster, Gini factor (morphGiniic)
719-728 F10.6 --- M20ic ?=-99 Morphological features for the initial
cluster, normalized second-order moment of
the 20 per cent of brightest pixels of the
galaxy (morphM20ic)
730-739 F10.6 --- C1ic ?=-99 Morphological features for the initial
cluster, concentration index C1
(C1 = log(r80/r20) (morphconc1_ic)
741-750 F10.6 --- C2ic ?=-99 Morphological features for the initial
cluster, concentration index C2
(C2 = log(r90/r50) (morphconc2_ic)
752-760 F9.5 --- clumpic ?=-99 Morphological features for the initial
cluster, clumpiness index
(morphclumpinessic)
762-771 F10.6 --- rP/rmic ?=-99 Morphological features for the initial
cluster, ratio of the Petrosian radius to the
deprojected source cluster rmax
(morphrPetrosiantormaxic)
773-782 F10.6 --- asymmic ?=-99 Morphological features for the initial
cluster, asymmetry index (morphasymmic)
784-793 F10.6 --- r20/rmic ?=-99 Morphological features for the initial
cluster, ratio of the r20 to rmax
(morphr20tormaxic)
795-804 F10.6 --- r80/rmic ?=-99 Morphological features for the initial
cluster, ratio of the r80 to rmax
(morphr80tormaxic)
806-814 F9.5 --- rPetroic ?=-99 Morphological features for the initial
cluster,Petrosian radius
(morphrPetrosian_ic)
816-825 F10.6 --- Ginidepr ?=-99 Morphological features for the
deprojected cluster, Gini factor
(morphGinidepr)
827-836 F10.6 --- M20depr ?=-99 Morphological features for the
deprojected cluster, normalized second-order
moment of the 20 per cent of brightest pixels
of the galaxy (morphM20depr)
838-847 F10.6 --- C1depr ?=-99 Morphological features for the
deprojected cluster, concentration index C1
(C1 = log(r80/r20) (morphconc1_depr)
849-858 F10.6 --- C2depr ?=-99 Morphological features for the
deprojected cluster, concentration index
C2 (C2 = log(r90/r50) (morphconc2_depr)
860-868 F9.5 --- clumpdepr ?=-99 Morphological features for the
deprojected cluster, clumpiness index
(morphclumpinessdepr)
870-879 F10.6 --- rP/rmdepr ?=-99 Morphological features for the
deprojected cluster, ratio of the Petrosian
radius to the deprojected source cluster rmax
(morphrPetrosiantormaxdepr)
881-890 F10.6 --- asymmdepr ?=-99 Morphological features for the
deprojected cluster, asymmetry index
(morphasymmdepr)
892-901 F10.6 --- r20/rmdepr ?=-99 Morphological features for the
deprojected cluster, ratio of the r20 to
rmax (morphr20tormaxdepr)
903-912 F10.6 --- r80/rmdepr ?=-99 Morphological features for the
deprojected cluster, ratio of the r80 to
rmax (morphr80tormaxdepr)
914-922 F9.5 --- rPetrdepr ?=-99 Morphological features for the
deprojected cluster, Petrosian radius
(morphrPetrosian_depr)
924-933 F10.6 --- Giniunsh ?=-99 Morphological features for the unsharp
cluster, Gini factor (morphGiniunsh)
935-945 F11.6 --- M20unsh ?=-99 Morphological features for the unsharp
cluster, normalized second-order moment of
the 20 per cent of brightest pixels of the
galaxy (morphM20unsh)
947-956 F10.6 --- C1unsh ?=-99 Morphological features for the unsharp
cluster, concentration index C1
(C1 = log(r80/r20) (morphconc1_unsh)
958-967 F10.6 --- C2unsh ?=-99 Morphological features for the unsharp
cluster, concentration index C2
(C2 = log(r90/r50) (morphconc2_unsh)
969-978 F10.6 --- clumpunsh ?=-99 Morphological features for the unsharp
cluster, clumpiness index
(morphclumpinessunsh)
980-989 F10.6 --- rP/rmunsh ?=-99 Morphological features for the unsharp
cluster, ratio of the Petrosian radius to the
deprojected source cluster rmax
(morphrPetrosiantormaxunsh)
991-1000 F10.6 --- asymmunsh ?=-99 Morphological features for the unsharp
cluster, asymmetry index (morphasymmunsh)
1002-1011 F10.6 --- r20/rmunsh ?=-99 Morphological features for the unsharp
cluster, ratio of the r20 to rmax
(morphr20tormaxunsh)
1013-1022 F10.6 --- r80/rmunsh ?=-99 Morphological features for the unsharp
cluster, ratio of the r80 to rmax
(morphr80tormaxunsh)
1024-1032 F9.5 --- rPetrounsh ?=-99 Morphological features for the unsharp
cluster, Petrosian radius
(morphrPetrosian_unsh)
1034-1042 F9.6 --- rdancric ?=-1 Radial distribution features for the
density attractors, ratio of number non-core
density attractors to the total number of
points in the source cluster
(rdistrattrnoncoreratioic)
1044-1052 F9.6 --- rdamic ?=-1 Radial distribution features for the
density attractors, mode of the radial
distribution (rdistrattrmodeic)
1054-1062 F9.6 --- rdasigic ?=-1 Radial distribution features for the
density attractors, standard deviation of the
radial distribution (rdistrattrsigic)
1064-1072 F9.6 --- rdaskewic Radial distribution features for the density
attractors, skewness of the radial
distribution (rdistrattrskewic)
1074-1082 F9.6 --- rdskewpic ?=-1 Radial distribution features for the
density attractors, two-sided p-value for the
skewness test (rdistrattrskewpvalic)
1084-1093 F10.6 --- rdakurtic Radial distribution features for the density
attractors, kurtosis of the radial
distribution (rdistrattrkurtic)
1095-1103 F9.6 --- rdadrnzic ?=-1 Radial distribution features for the
density attractors, difference between the
largest and the smallest values of r, where
the distribution is larger than zero
(rdistrattrdeltarnonzero_ic)
1105-1114 F10.6 --- tdasigic ?=-1 Angular distribution features for the
density attractors, standard deviation of the
radial distribution (thetadistrattrsigic)
1116-1124 F9.6 --- tdaskewic Angular distribution features for the density
attractors, skewness of the radial
distribution (thetadistrattrskewic)
1126-1134 F9.6 --- tdaskewpc ?=-1 Angular distribution features for the
density attractors, two-sided p-value for the
skewness test
(thetadistrattrskewpvalic)
1136-1145 F10.6 --- tdakurtic Angular distribution features for the density
attractors, kurtosis of the radial
distribution (thetadistrattrkurtic)
1147-1156 F10.6 --- tdadtnzic ?=-1 Angular distribution features for the
density attractors, difference between the
largest and the smallest values of r, where
the distribution is larger than zero
(thetadistrattrdeltathetanonzero_ic)
--------------------------------------------------------------------------------
Note (1): GZ class label as follows:
E = elliptical
S = spiral
Note (2): Label for task 01 of the GZ2 decision tree as follows:
A1 = smooth
A2 = feature or disc
Note (3): Label for task 04 of the GZ2 decision tree as follows:
A1 = this could be a disc
A2 = this oculd not be a disc
NO = no decision
Note (4): if negative, it means no source was detected, or source detection
failure.
--------------------------------------------------------------------------------
History:
From electronic version of the journal
(End) Patricia Vannier [CDS] 14-May-2018