grup de tecnologies digitals grup de tecnologies digitals departament de tecnologies digitals i de departament de tecnologies digitals i de escola politècnica superior escola politècnica superior 2D and 3D image surface descriptors for fish otoliths classification authors: Ramon Reig i Bolaño (UVIC) Pere Martí i Puig (UVIC) Antoni Lombarte (ICM- CSIC) Vicenç Parisi i Baradad (UPC)
34
Embed
2D and 3D image surface descriptors for fish otoliths classification authors:Ramon Reig i Bolaño (UVIC) Pere Martí i Puig (UVIC) Antoni Lombarte (ICM-CSIC)
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
grup de tecnologies digitalsgrup de tecnologies digitalsdepartament de tecnologies digitals i de la informaciódepartament de tecnologies digitals i de la informacióescola politècnica superiorescola politècnica superior
2D and 3D image surface descriptors for fish otoliths classification
authors: Ramon Reig i Bolaño (UVIC)Pere Martí i Puig (UVIC)Antoni Lombarte (ICM-CSIC)Vicenç Parisi i Baradad (UPC)
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
2/32
outline of the talk
introduction: fish otoliths, AFORO database
2D contour descriptors
new 2D contour descriptors
3D surfaces of otholits: proposal of extension
2nd. order descriptors (UDWT)
conclusions
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
3/32introduction:morphological description of fish otoliths
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
4/32
introduction: AFORO database
ICM collectionassembles image otoliths, morphometrics and shape analysis
2456 high resolution images, 776 species and 156 families
AFORO: shape analysis of fish otolithsLombarte, A., Ò. Chic, V. Parisi-Baradad, R. Olivella, J. Piera & E. García-Ladona. 2006. A web-based environment from shape analysis of fish otoliths. The AFORO database. Scientia Marina 70: 147-152.
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
5/32
outline of the talk
introduction: fish otoliths, AFORO database
2D contour descriptors
new 2D contour descriptors
3D surfaces of otholits: proposal of extension
2nd. order descriptors (UDWT)
conclusions
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
6/32
2D contour descriptors
100 200 300 400 500 600
100
150
200
250
300
350
200 400 600 800 1000 1200 1400
100
200
300
y component200 400 600 800 1000 1200 1400
100
200
300
400
500
600
x component
high quality imagebinarized image contour
components xy
lossy equiangular polar code lossless chain code
CSS, curvature scale spacewavelet analysis
1st order
2nd order
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
7/32
robust to affine transformation and origin
lossy: non perfect reconstruction from polar to xy coordinates / contour
1st. order: equiangular radial values
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
8/32
outline of the talk
introduction: fish otoliths, AFORO database
2D contour descriptors
new 2D contour descriptors
3D surfaces of otholits: proposal of extension
2nd. order descriptors (UDWT)
conclusions
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
9/32
new 2D contour descriptors
100 200 300 400 500 600
100
150
200
250
300
350
200 400 600 800 1000 1200 1400
100
200
300
y component200 400 600 800 1000 1200 1400
100
200
300
400
500
600
x component
high quality image
binarized image contourcomponents xy
lossy direct sampled y code
reflected contour image
200 400 600 800 1000 1200
100
200
300
100 200 300 400 500 600 700 800 900 1000 1100
100
200
300
direct sample y
200 400 600 800 1000 1200 14000
2
4
6
chain code
200 400 600 800 1000 1200 1400
4
6
8
10
12modified chain codelossless modified chain code 1st order
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
10/32
binary contour reflexion
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
11/32
non redundant contour
200 400 600 800 1000 1200
100
200
300
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
14/32
200 400 600 800 1000 1200 14000
2
4
6
chain code
200 400 600 800 1000 1200 1400
4
6
8
10
12modified chain code
200 400 600 800 1000 1200
100
200
300
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
15/32
100 200 300 400 500 600 700 800 900 1000 1100
100
200
300
direct sample y
200 400 600 800 1000 1200
100
200
300
540 560 580 600
140
160
180
200
220
240
direct sample y
1st. order lossy coder: direct sampled y
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
16/32
contour reconstruction (lossless vs. lossy)
100 200 300 400 500 600
100
150
200
250
300
350
new contour approximation
chain code
direct sample
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
17/32
outline of the talk
introduction: fish otoliths, AFORO database
2D contour descriptors
new 2D contour descriptors
3D surfaces of otholits: proposal of extension
2nd. order descriptors (UDWT)
conclusions
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
18/32
3D surfaces of otholits
volumetric 3D surface much more details for the classification/description process
• (i.e. the representation of the acoustic groove)
normalized 3D data adquisition and orientation of otolith
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
19/32
proposal of extension to 3D surfaces
high quality volume surface volume surface reflected on zy plane
lossy sampled surface, image on xyx
z
y
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
20/32
proposal of extension to 3D surfaces
high quality volume surface volume surface reflected on yx plane
lossy sampled surface, image on zx z
yx
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
21/32
outline of the talk
introduction: fish otoliths, AFORO database
2D contour descriptors
new 2D contour descriptors
3D surfaces of otholits: proposal of extension
2nd. order descriptors (UDWT)
conclusions
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
22/32
HPa1
v vd1
LPa1
va1HPa2
vd2
LPa2
va2HPa3
LPa3 ….
vd3
va3
2nd. order: UDWT (undecimated wavelet transf.)
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
0
10
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
0
10
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
0
10
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
0
10
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
0
10
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
23/32
wavelet analysis of lossy sampled y code
approximate (LP) details wav. (HP)
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100
-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-505
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
0
10
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
100 200 300 400 500 600 700 800 900 1000 1100
100200300
100 200 300 400 500 600 700 800 900 1000 1100-10
010
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
24/32
wavelet analysis of lossless modified chain code
approximate (LP) details wav. (HP)
200 400 600 800 1000 1200 1400
5
10
200 400 600 800 1000 1200 14005
10
200 400 600 800 1000 1200 1400-101
200 400 600 800 1000 1200 14005
10
200 400 600 800 1000 1200 1400
-0.50
0.51
200 400 600 800 1000 1200 140068
10
200 400 600 800 1000 1200 1400
-0.50
0.51
200 400 600 800 1000 1200 140068
10
200 400 600 800 1000 1200 1400-101
200 400 600 800 1000 1200 14006
8
200 400 600 800 1000 1200 1400-1
-0.50
0.5
200 400 600 800 1000 1200 1400
789
200 400 600 800 1000 1200 1400-0.5
00.5
200 400 600 800 1000 1200 1400789
200 400 600 800 1000 1200 1400-0.4-0.2
00.20.4
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
25/32
outline of the talk
introduction: fish otoliths, AFORO database
2D contour descriptors
new 2D contour descriptors
3D surfaces of otholits: proposal of extension
2nd. order descriptors (UDWT)
conclusions & questions
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
26/32
conclusions
we proposed a new lossy 2D contour coder (sampled y code) and a lossless one (modified chain code):
based on reflectionbased on reflection
extensible to 3D surfaces (extensible to 3D surfaces (lossy sampled y codelossy sampled y code))
can use 2nd. order descriptorscan use 2nd. order descriptors
drawback: need a normalized adquisition and positioning,
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
27/32
questions?
thank you
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
28/32
DTW (dynamic time warping), alignment
20 40 60 807
8
9
10
11
20 40 60 808
9
10
20 40 60 80-1
0
1
20 40 60 80
8.5
9
9.5
20 40 60 80
-0.4
-0.2
0
0.2
0.4
20 40 60 80
8.5
9
9.5
20 40 60 80-0.4
-0.2
0
0.2
10 20 30 40 50 60 70 80 90
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
29/32
accumulative differences matrix
minimum distance path, backwards
njniWnnjiopt ,,,minarg:2,
DARDARDA
L,1 ,,min
..., , 2
jiDARDARDARDDAR
D
rsr n
1,ji1i,j11,jii,ji,j
sri,j
rrr
ji
n1
DTW (dynamic time warping) algorithm
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
30/32
10 20 30 40 50 60 70 80 90
10
20
30
40
50
60
70
80
90
10 20 30 40 50 60 70 80 90
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
10
20
30
40
50
60
70
80
90
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
31/32
0 0.5 10
0.2
0.4
0.6
0.8
1Analysis until level 2
0 0.5 10
0.2
0.4
0.6
0.8
1Synthesis from level 2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
0.2
0.4
0.6
0.8
1
Global Composition level 2
filterbank
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
32/32
50 100 150 200 250 300 350
-0.5
0
0.5
Correlacio Aproximacions ordre 3 bsel = 0
0 50 100 150 200 250 300 350 4007
8
9
10
measure of MCC to find similar descriptors
N
mknymnxkr
rr
krk
n yxxy
yyxx
xyxy
00
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
33/32
introduction: AFORO database
AFORO: shape analysis of fish otolithsLombarte, A., Ò. Chic, V. Parisi-Baradad, R. Olivella, J. Piera & E. García-Ladona. 2006. A web-based environment from shape analysis of fish otoliths. The AFORO database. Scientia Marina 70: 147-152.
2456 high resolution images, 776 species and 156 families
morphometry
shape analysis contours:
• FT (Fourier transform)
• CSS (curvature space scale)
• WT (wavelet analysis)
authomatic taxonomy identification
2D and 3D image surface descriptors for fish otoliths classification – 2nd workshop in signal processing for marine and seismic data, 2009
34/32
robust to affine transformation and origin
lossless: perfect reconstruction to xy coordinates / contour