SlideShare une entreprise Scribd logo
1  sur  17
Télécharger pour lire hors ligne
Binary Symbol Recognition from Local Dissimilarity Map
                             GREC - 2009


                 F. Morain-Nicolier, J. Landr´, S. Ruan
                                             e

                     frederic.nicolier@univ-reims.fr
                         http://pixel-shaker.fr

                        CRESTIC - URCA/IUT Troyes


                            22 juillet 2009



1
Goal




       Symbol Recognition using raw pixel information

       Not a statistical approach : no learning
       Not a structural approach : no primitive

       ⇒ Build a good (di)similarity measure between images




2                                                             2/14
Local to Global Dissimilarity
       Local Dissimilarity Map

                        LDMA,B   =   |A − B| max(dtA , dt(p))   (1)
                                 =   BdtA + AdtB .              (2)




3                                                                 3/14
Local to Global Dissimilarity
       Local Dissimilarity Map

                        LDMA,B    =    |A − B| max(dtA , dt(p))            (1)
                                  =    BdtA + AdtB .                       (2)




       Global Dissimilarity Measure : sum the (squared) values :

                 GDM(A, B) = α        B(p)dt2 (p) + β
                                            A               A(p)dt2 (p).
                                                                  B        (3)
                                  p                     p
4                                                                            3/14
Local to Global Dissimilarity


       Why α and β ?
       Chamfer Matching [Borgefors] : sum of the distances of each translated
       model’s pixel to the nearest images’s pixel.
       GDM is a double Chamfer Matching :

                           GDMA,B ∼ αCS(A, B) + βCS(B, A)                       (4)

       CS : how is M similar to I ?
       GDM : + how is I similar to M ?
       symetric or asymetric matching (links to psychological models of similarity
       [Tversky]) ?



5                                                                                 4/14
First Algorithm



     1   I contains an unknow symbol.
     2   Foreach model M :
             compute the dissimilarity GDM(I , M) between I and M.
     3   Keep the model with the lowest dissimilarity as winner.




6                                                                    5/14
First Algorithm



     1   I contains an unknow symbol.
     2   Foreach model M :
             compute the dissimilarity GDM(I , M) between I and M.
     3   Keep the model with the lowest dissimilarity as winner.

         Tested on the GREC2005 international symbol recognition contest database
         (electronic and architecture) : six degradations
         25 symbols to be recognized in 50 images.
         α = 1 and β = 0.




7                                                                                   5/14
Results (no deformation)

                           Degradation 1




                               100 %
8                                          6/14
Results (no deformation)

                           Degradation 2




                               100 %
9                                          7/14
Results (no deformation)

                        Degradation 3




                            100 %
10                                      8/14
Results (no deformation)

                        Degradation 4




                            100 %
11                                      9/14
Results (no deformation)

                        Degradation 5




                            100 %
12                                      10/14
Results (no deformation)

                          Degradation 6




                    54 % (but best is 59.81 %)
13                                               11/14
And with deformations ?



     Transform rotation and scale into translations
     ⇒ log-polar transform




14                                                    12/14
And with deformations ?



     Transform rotation and scale into translations
     ⇒ log-polar transform
     Given two images I and M to be registrated :
         Compute Ilp and Mlp
         Which translation produces the most similar images ?
         ⇒ compute the global dississimilarity between Ilp and each translation of Mlp .
         Very fast with cross-correlations (in Fourier Domain) :

                                LocI ,M = αdt2
                                             I   M + βI    dt2 .
                                                             M                       (5)

         Find the minimum of LocI ,M and find the rotation and scale parameters from
         its position.




15                                                                                     12/14
Complete Algorithm



     1   Compute Ilp , log-polar representation of I .
     2   Foreach model M :
           1   compute Mlp , log-polar representation of M
           2   find the global minimum of LocIlp ,Mlp (fast computation)
           3   find (σ, θ) the scaling and rotation parameters of M.
           4   compute Mcorrected from (ρ, θ), by applying reverse scale and rotation.
           5   compute the dissimilarity GDM(I , Mcorrected ) between Mcorrected and I .
     3   Keep the model with the lowest dissimilarity as winner.




16                                                                                         13/14
Advantages and Drawbacks


           rot/scale     m1       m2       m3       m4       m5      m6
            without    100%     100%     100%     100%     100%     54%
             with       96%      40%      94%      70%      42%     12%

     Quite good performances (lowered by 5%-10% with 100 images/150
     symbols)
     without any a-priori knowledge, pre-processing, nor primitive extraction
     Fast but still needs parameters estimations
     Next :
         Gray level images ⇒ Localization with rotation/scale variations.
         Balance the asymetry, e.g. α = 0.9 and β = 0.1 (take into account noise)
         Direct measure in log-polar domain.
         Apply preprocessing to boost performances.


17                                                                                  14/14

Contenu connexe

Tendances

Dilation and erosion
Dilation and erosionDilation and erosion
Dilation and erosion
Aswin Pv
 
Hidden surface removal
Hidden surface removalHidden surface removal
Hidden surface removal
Ankit Garg
 

Tendances (20)

Boundary Extraction
Boundary ExtractionBoundary Extraction
Boundary Extraction
 
Morphological image processing
Morphological image processingMorphological image processing
Morphological image processing
 
Ce 2009 Gate Paper Prsolutions08
Ce 2009 Gate Paper Prsolutions08Ce 2009 Gate Paper Prsolutions08
Ce 2009 Gate Paper Prsolutions08
 
Region filling
Region fillingRegion filling
Region filling
 
Morphology in graphics and image processing
Morphology in graphics and image processingMorphology in graphics and image processing
Morphology in graphics and image processing
 
A primer for colour computer vision
A primer for colour computer visionA primer for colour computer vision
A primer for colour computer vision
 
Hit and-miss transform
Hit and-miss transformHit and-miss transform
Hit and-miss transform
 
Morphological operations
Morphological operationsMorphological operations
Morphological operations
 
Digital signal and image processing FAQ
Digital signal and image processing FAQDigital signal and image processing FAQ
Digital signal and image processing FAQ
 
morphological image processing
morphological image processingmorphological image processing
morphological image processing
 
Morphological Image Processing
Morphological Image ProcessingMorphological Image Processing
Morphological Image Processing
 
Morphological image processing
Morphological image processingMorphological image processing
Morphological image processing
 
COM2304: Morphological Image Processing
COM2304: Morphological Image ProcessingCOM2304: Morphological Image Processing
COM2304: Morphological Image Processing
 
SURF - Speeded Up Robust Features
SURF - Speeded Up Robust FeaturesSURF - Speeded Up Robust Features
SURF - Speeded Up Robust Features
 
Dilation and erosion
Dilation and erosionDilation and erosion
Dilation and erosion
 
Gch12 l1
Gch12 l1Gch12 l1
Gch12 l1
 
Hidden surface removal
Hidden surface removalHidden surface removal
Hidden surface removal
 
Implement the morphological operations: Dilation, Erosion, Opening and Closing
Implement the morphological operations: Dilation, Erosion, Opening and ClosingImplement the morphological operations: Dilation, Erosion, Opening and Closing
Implement the morphological operations: Dilation, Erosion, Opening and Closing
 
Lecture 15 image morphology examples
Lecture 15 image morphology examplesLecture 15 image morphology examples
Lecture 15 image morphology examples
 
Hidden surface removal algorithm
Hidden surface removal algorithmHidden surface removal algorithm
Hidden surface removal algorithm
 

En vedette

Mp26 : Connecting Startups with Talents
Mp26 : Connecting Startups with TalentsMp26 : Connecting Startups with Talents
Mp26 : Connecting Startups with Talents
Montreal Python
 
ENGLISH FOR ALL
ENGLISH FOR ALLENGLISH FOR ALL
ENGLISH FOR ALL
sacha
 
Mp25: Optical Music Recognition with Python
Mp25: Optical Music Recognition with PythonMp25: Optical Music Recognition with Python
Mp25: Optical Music Recognition with Python
Montreal Python
 
Music Notation
Music NotationMusic Notation
Music Notation
jansonke
 
Music recognition
Music recognition Music recognition
Music recognition
aaronloklok
 
Petit Club Social TV - Shazam presentation par Iain Dendle
Petit Club Social TV - Shazam presentation par Iain DendlePetit Club Social TV - Shazam presentation par Iain Dendle
Petit Club Social TV - Shazam presentation par Iain Dendle
Petit Web
 
The importance of role specific training for
The importance of role specific training forThe importance of role specific training for
The importance of role specific training for
Paul Sandell
 

En vedette (20)

Mp26 : Connecting Startups with Talents
Mp26 : Connecting Startups with TalentsMp26 : Connecting Startups with Talents
Mp26 : Connecting Startups with Talents
 
Representation in Music Videos
Representation in Music VideosRepresentation in Music Videos
Representation in Music Videos
 
Welcome to Music 101
Welcome to Music 101Welcome to Music 101
Welcome to Music 101
 
Music terminology
Music terminologyMusic terminology
Music terminology
 
Representation [RE UPLOAD]
Representation [RE UPLOAD]Representation [RE UPLOAD]
Representation [RE UPLOAD]
 
ENGLISH FOR ALL
ENGLISH FOR ALLENGLISH FOR ALL
ENGLISH FOR ALL
 
Mp25: Optical Music Recognition with Python
Mp25: Optical Music Recognition with PythonMp25: Optical Music Recognition with Python
Mp25: Optical Music Recognition with Python
 
Welcome To Music 101
Welcome To Music 101Welcome To Music 101
Welcome To Music 101
 
Gracenote Music Recognition, Metadata, and Discovery APIs
Gracenote Music Recognition, Metadata, and Discovery APIsGracenote Music Recognition, Metadata, and Discovery APIs
Gracenote Music Recognition, Metadata, and Discovery APIs
 
MM Live! Shazam
MM Live! Shazam MM Live! Shazam
MM Live! Shazam
 
Music Notation
Music NotationMusic Notation
Music Notation
 
Presentation_final
Presentation_finalPresentation_final
Presentation_final
 
Music recognition
Music recognition Music recognition
Music recognition
 
Petit Club Social TV - Shazam presentation par Iain Dendle
Petit Club Social TV - Shazam presentation par Iain DendlePetit Club Social TV - Shazam presentation par Iain Dendle
Petit Club Social TV - Shazam presentation par Iain Dendle
 
Understanding parts of a music staff
Understanding parts of a music staffUnderstanding parts of a music staff
Understanding parts of a music staff
 
Reading and writing music
Reading and writing musicReading and writing music
Reading and writing music
 
The importance of role specific training for
The importance of role specific training forThe importance of role specific training for
The importance of role specific training for
 
What does the word “classical” mean
What does the word “classical” meanWhat does the word “classical” mean
What does the word “classical” mean
 
Music Jeopardy
Music JeopardyMusic Jeopardy
Music Jeopardy
 
Optical Mark Recognition
Optical Mark RecognitionOptical Mark Recognition
Optical Mark Recognition
 

Similaire à Binary Symbol Recognition from Local Dissimilarity Map

Bi-layer segmentation of binocular stereo video
Bi-layer segmentation of binocular stereo videoBi-layer segmentation of binocular stereo video
Bi-layer segmentation of binocular stereo video
Videoguy
 

Similaire à Binary Symbol Recognition from Local Dissimilarity Map (11)

Fingerprint High Level Classification
Fingerprint High Level ClassificationFingerprint High Level Classification
Fingerprint High Level Classification
 
Facial Expression Recognition / Removal
Facial Expression Recognition / RemovalFacial Expression Recognition / Removal
Facial Expression Recognition / Removal
 
Muzammil Abdulrahman PPT On Gabor Wavelet Transform (GWT) Based Facial Expres...
Muzammil Abdulrahman PPT On Gabor Wavelet Transform (GWT) Based Facial Expres...Muzammil Abdulrahman PPT On Gabor Wavelet Transform (GWT) Based Facial Expres...
Muzammil Abdulrahman PPT On Gabor Wavelet Transform (GWT) Based Facial Expres...
 
=SLAM ppt.pdf
=SLAM ppt.pdf=SLAM ppt.pdf
=SLAM ppt.pdf
 
Copy-Rotate-Move Forgery Detection Based on Spatial Domain
Copy-Rotate-Move Forgery Detection Based on Spatial DomainCopy-Rotate-Move Forgery Detection Based on Spatial Domain
Copy-Rotate-Move Forgery Detection Based on Spatial Domain
 
Bi-layer segmentation of binocular stereo video
Bi-layer segmentation of binocular stereo videoBi-layer segmentation of binocular stereo video
Bi-layer segmentation of binocular stereo video
 
Representing Simplicial Complexes with Mangroves
Representing Simplicial Complexes with MangrovesRepresenting Simplicial Complexes with Mangroves
Representing Simplicial Complexes with Mangroves
 
Face Alignment Using Active Shape Model And Support Vector Machine
Face Alignment Using Active Shape Model And Support Vector MachineFace Alignment Using Active Shape Model And Support Vector Machine
Face Alignment Using Active Shape Model And Support Vector Machine
 
Model Care96
Model Care96Model Care96
Model Care96
 
Reweighting and Boosting to uniforimty in HEP
Reweighting and Boosting to uniforimty in HEPReweighting and Boosting to uniforimty in HEP
Reweighting and Boosting to uniforimty in HEP
 
Standardizing inter-element distances in repertory grids
Standardizing inter-element distances in repertory gridsStandardizing inter-element distances in repertory grids
Standardizing inter-element distances in repertory grids
 

Plus de Frédéric Morain-Nicolier

2012.09.25 - Local and non-metric similarities between images - why, how and ...
2012.09.25 - Local and non-metric similarities between images - why, how and ...2012.09.25 - Local and non-metric similarities between images - why, how and ...
2012.09.25 - Local and non-metric similarities between images - why, how and ...
Frédéric Morain-Nicolier
 
Mesures de dissimilarités locales et globales entre images, symétriques et as...
Mesures de dissimilarités locales et globales entre images, symétriques et as...Mesures de dissimilarités locales et globales entre images, symétriques et as...
Mesures de dissimilarités locales et globales entre images, symétriques et as...
Frédéric Morain-Nicolier
 

Plus de Frédéric Morain-Nicolier (14)

Practising Fourier Analysis with Digital Images
Practising Fourier Analysis with Digital ImagesPractising Fourier Analysis with Digital Images
Practising Fourier Analysis with Digital Images
 
GEII - Ma3 - Matrices
GEII - Ma3 - MatricesGEII - Ma3 - Matrices
GEII - Ma3 - Matrices
 
GEII - Ma3 - Suites et séries
GEII - Ma3 - Suites et sériesGEII - Ma3 - Suites et séries
GEII - Ma3 - Suites et séries
 
GEII - Ma3 - Représentations de Fourier et convolution
GEII - Ma3 - Représentations de Fourier et convolutionGEII - Ma3 - Représentations de Fourier et convolution
GEII - Ma3 - Représentations de Fourier et convolution
 
GEII - OL3 : Signaux et systèmes numériques
GEII - OL3 : Signaux et systèmes numériquesGEII - OL3 : Signaux et systèmes numériques
GEII - OL3 : Signaux et systèmes numériques
 
Étude des fonctions à plusieurs variables (GEII MA32)
Étude des fonctions à plusieurs variables (GEII MA32)Étude des fonctions à plusieurs variables (GEII MA32)
Étude des fonctions à plusieurs variables (GEII MA32)
 
Introduction à la transformée en z et convolution discrète (GEII MA32)
Introduction à la transformée en z et convolution discrète (GEII MA32)Introduction à la transformée en z et convolution discrète (GEII MA32)
Introduction à la transformée en z et convolution discrète (GEII MA32)
 
2012.09.25 - Local and non-metric similarities between images - why, how and ...
2012.09.25 - Local and non-metric similarities between images - why, how and ...2012.09.25 - Local and non-metric similarities between images - why, how and ...
2012.09.25 - Local and non-metric similarities between images - why, how and ...
 
Ipta2010
Ipta2010Ipta2010
Ipta2010
 
Capteurs de positions
Capteurs de positionsCapteurs de positions
Capteurs de positions
 
CCD
CCDCCD
CCD
 
SRC - SAV - La lumière
SRC - SAV - La lumièreSRC - SAV - La lumière
SRC - SAV - La lumière
 
SRC - SAV - Système Visuel et Auditif Humain
SRC - SAV - Système Visuel et Auditif HumainSRC - SAV - Système Visuel et Auditif Humain
SRC - SAV - Système Visuel et Auditif Humain
 
Mesures de dissimilarités locales et globales entre images, symétriques et as...
Mesures de dissimilarités locales et globales entre images, symétriques et as...Mesures de dissimilarités locales et globales entre images, symétriques et as...
Mesures de dissimilarités locales et globales entre images, symétriques et as...
 

Dernier

Jual Obat Aborsi Hongkong ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
Jual Obat Aborsi Hongkong ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...Jual Obat Aborsi Hongkong ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
Jual Obat Aborsi Hongkong ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
ZurliaSoop
 
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in DelhiRussian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
kauryashika82
 
The basics of sentences session 3pptx.pptx
The basics of sentences session 3pptx.pptxThe basics of sentences session 3pptx.pptx
The basics of sentences session 3pptx.pptx
heathfieldcps1
 
Spellings Wk 3 English CAPS CARES Please Practise
Spellings Wk 3 English CAPS CARES Please PractiseSpellings Wk 3 English CAPS CARES Please Practise
Spellings Wk 3 English CAPS CARES Please Practise
AnaAcapella
 

Dernier (20)

Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
Explore beautiful and ugly buildings. Mathematics helps us create beautiful d...
 
Jual Obat Aborsi Hongkong ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
Jual Obat Aborsi Hongkong ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...Jual Obat Aborsi Hongkong ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
Jual Obat Aborsi Hongkong ( Asli No.1 ) 085657271886 Obat Penggugur Kandungan...
 
Mixin Classes in Odoo 17 How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17  How to Extend Models Using Mixin ClassesMixin Classes in Odoo 17  How to Extend Models Using Mixin Classes
Mixin Classes in Odoo 17 How to Extend Models Using Mixin Classes
 
Python Notes for mca i year students osmania university.docx
Python Notes for mca i year students osmania university.docxPython Notes for mca i year students osmania university.docx
Python Notes for mca i year students osmania university.docx
 
ICT role in 21st century education and it's challenges.
ICT role in 21st century education and it's challenges.ICT role in 21st century education and it's challenges.
ICT role in 21st century education and it's challenges.
 
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptxSKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
SKILL OF INTRODUCING THE LESSON MICRO SKILLS.pptx
 
Making communications land - Are they received and understood as intended? we...
Making communications land - Are they received and understood as intended? we...Making communications land - Are they received and understood as intended? we...
Making communications land - Are they received and understood as intended? we...
 
psychiatric nursing HISTORY COLLECTION .docx
psychiatric  nursing HISTORY  COLLECTION  .docxpsychiatric  nursing HISTORY  COLLECTION  .docx
psychiatric nursing HISTORY COLLECTION .docx
 
Accessible Digital Futures project (20/03/2024)
Accessible Digital Futures project (20/03/2024)Accessible Digital Futures project (20/03/2024)
Accessible Digital Futures project (20/03/2024)
 
Understanding Accommodations and Modifications
Understanding  Accommodations and ModificationsUnderstanding  Accommodations and Modifications
Understanding Accommodations and Modifications
 
ComPTIA Overview | Comptia Security+ Book SY0-701
ComPTIA Overview | Comptia Security+ Book SY0-701ComPTIA Overview | Comptia Security+ Book SY0-701
ComPTIA Overview | Comptia Security+ Book SY0-701
 
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in DelhiRussian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
Russian Escort Service in Delhi 11k Hotel Foreigner Russian Call Girls in Delhi
 
Unit-IV- Pharma. Marketing Channels.pptx
Unit-IV- Pharma. Marketing Channels.pptxUnit-IV- Pharma. Marketing Channels.pptx
Unit-IV- Pharma. Marketing Channels.pptx
 
General Principles of Intellectual Property: Concepts of Intellectual Proper...
General Principles of Intellectual Property: Concepts of Intellectual  Proper...General Principles of Intellectual Property: Concepts of Intellectual  Proper...
General Principles of Intellectual Property: Concepts of Intellectual Proper...
 
Unit-V; Pricing (Pharma Marketing Management).pptx
Unit-V; Pricing (Pharma Marketing Management).pptxUnit-V; Pricing (Pharma Marketing Management).pptx
Unit-V; Pricing (Pharma Marketing Management).pptx
 
The basics of sentences session 3pptx.pptx
The basics of sentences session 3pptx.pptxThe basics of sentences session 3pptx.pptx
The basics of sentences session 3pptx.pptx
 
Key note speaker Neum_Admir Softic_ENG.pdf
Key note speaker Neum_Admir Softic_ENG.pdfKey note speaker Neum_Admir Softic_ENG.pdf
Key note speaker Neum_Admir Softic_ENG.pdf
 
Third Battle of Panipat detailed notes.pptx
Third Battle of Panipat detailed notes.pptxThird Battle of Panipat detailed notes.pptx
Third Battle of Panipat detailed notes.pptx
 
Spellings Wk 3 English CAPS CARES Please Practise
Spellings Wk 3 English CAPS CARES Please PractiseSpellings Wk 3 English CAPS CARES Please Practise
Spellings Wk 3 English CAPS CARES Please Practise
 
Kodo Millet PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
Kodo Millet  PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...Kodo Millet  PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
Kodo Millet PPT made by Ghanshyam bairwa college of Agriculture kumher bhara...
 

Binary Symbol Recognition from Local Dissimilarity Map

  • 1. Binary Symbol Recognition from Local Dissimilarity Map GREC - 2009 F. Morain-Nicolier, J. Landr´, S. Ruan e frederic.nicolier@univ-reims.fr http://pixel-shaker.fr CRESTIC - URCA/IUT Troyes 22 juillet 2009 1
  • 2. Goal Symbol Recognition using raw pixel information Not a statistical approach : no learning Not a structural approach : no primitive ⇒ Build a good (di)similarity measure between images 2 2/14
  • 3. Local to Global Dissimilarity Local Dissimilarity Map LDMA,B = |A − B| max(dtA , dt(p)) (1) = BdtA + AdtB . (2) 3 3/14
  • 4. Local to Global Dissimilarity Local Dissimilarity Map LDMA,B = |A − B| max(dtA , dt(p)) (1) = BdtA + AdtB . (2) Global Dissimilarity Measure : sum the (squared) values : GDM(A, B) = α B(p)dt2 (p) + β A A(p)dt2 (p). B (3) p p 4 3/14
  • 5. Local to Global Dissimilarity Why α and β ? Chamfer Matching [Borgefors] : sum of the distances of each translated model’s pixel to the nearest images’s pixel. GDM is a double Chamfer Matching : GDMA,B ∼ αCS(A, B) + βCS(B, A) (4) CS : how is M similar to I ? GDM : + how is I similar to M ? symetric or asymetric matching (links to psychological models of similarity [Tversky]) ? 5 4/14
  • 6. First Algorithm 1 I contains an unknow symbol. 2 Foreach model M : compute the dissimilarity GDM(I , M) between I and M. 3 Keep the model with the lowest dissimilarity as winner. 6 5/14
  • 7. First Algorithm 1 I contains an unknow symbol. 2 Foreach model M : compute the dissimilarity GDM(I , M) between I and M. 3 Keep the model with the lowest dissimilarity as winner. Tested on the GREC2005 international symbol recognition contest database (electronic and architecture) : six degradations 25 symbols to be recognized in 50 images. α = 1 and β = 0. 7 5/14
  • 8. Results (no deformation) Degradation 1 100 % 8 6/14
  • 9. Results (no deformation) Degradation 2 100 % 9 7/14
  • 10. Results (no deformation) Degradation 3 100 % 10 8/14
  • 11. Results (no deformation) Degradation 4 100 % 11 9/14
  • 12. Results (no deformation) Degradation 5 100 % 12 10/14
  • 13. Results (no deformation) Degradation 6 54 % (but best is 59.81 %) 13 11/14
  • 14. And with deformations ? Transform rotation and scale into translations ⇒ log-polar transform 14 12/14
  • 15. And with deformations ? Transform rotation and scale into translations ⇒ log-polar transform Given two images I and M to be registrated : Compute Ilp and Mlp Which translation produces the most similar images ? ⇒ compute the global dississimilarity between Ilp and each translation of Mlp . Very fast with cross-correlations (in Fourier Domain) : LocI ,M = αdt2 I M + βI dt2 . M (5) Find the minimum of LocI ,M and find the rotation and scale parameters from its position. 15 12/14
  • 16. Complete Algorithm 1 Compute Ilp , log-polar representation of I . 2 Foreach model M : 1 compute Mlp , log-polar representation of M 2 find the global minimum of LocIlp ,Mlp (fast computation) 3 find (σ, θ) the scaling and rotation parameters of M. 4 compute Mcorrected from (ρ, θ), by applying reverse scale and rotation. 5 compute the dissimilarity GDM(I , Mcorrected ) between Mcorrected and I . 3 Keep the model with the lowest dissimilarity as winner. 16 13/14
  • 17. Advantages and Drawbacks rot/scale m1 m2 m3 m4 m5 m6 without 100% 100% 100% 100% 100% 54% with 96% 40% 94% 70% 42% 12% Quite good performances (lowered by 5%-10% with 100 images/150 symbols) without any a-priori knowledge, pre-processing, nor primitive extraction Fast but still needs parameters estimations Next : Gray level images ⇒ Localization with rotation/scale variations. Balance the asymetry, e.g. α = 0.9 and β = 0.1 (take into account noise) Direct measure in log-polar domain. Apply preprocessing to boost performances. 17 14/14