Shape Matching - University of...
Transcript of Shape Matching - University of...
![Page 1: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/1.jpg)
Shape MatchingBrandon Smith and Shengnan Wang
Computer Vision – CS766 – Fall 2007
![Page 2: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/2.jpg)
Outline
• Introduction and Background
• Uses of shape matching
• Kinds of shape matching
• Support Vector Machine (SVM)
• Matching with Shape Contexts
• Shape Context
• Bipartite Graph Matching
• Modeling Transformations
• Invariance and Robustness
• Results
• Questions
• Shengnan’s part…
![Page 3: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/3.jpg)
Introduction and Background
Shape matching examples
Fruit InspectionFingerprint Matching
Hieroglyph Lookup Trademark Lookup
![Page 4: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/4.jpg)
Introduction and Background
![Page 5: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/5.jpg)
Introduction and Background
• Feature-Based Methods
• Brightness-Based Methods
![Page 6: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/6.jpg)
Introduction and Background
Feature-Based Methods
![Page 7: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/7.jpg)
Introduction and Background
Brightness-Based Methods
Two different frameworks:
• Explicitly find correspondences
• Build classifiers without explicitly finding
correspondences.
![Page 8: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/8.jpg)
Introduction and Background
Support Vector Machine (SVM)
![Page 9: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/9.jpg)
Introduction and Background
Approach:
1.Find correspondences between shapes
2.Estimate an aligning transform
3.Measure similarity
![Page 10: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/10.jpg)
Matching with Shape Contexts
Shape Context
![Page 11: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/11.jpg)
Matching with Shape Contexts
Shape Context
![Page 12: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/12.jpg)
Matching with Shape Contexts
Shape Context
3
9
![Page 13: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/13.jpg)
Matching with Shape Contexts
Shape Context
Belongie, et al, PAMI 2002, Shape matching and object recognition using shape contexts
![Page 14: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/14.jpg)
Matching with Shape Contexts
Shape Context
Belongie, et al, PAMI 2002, Shape matching and object recognition using shape contexts
K
k ji
ji
jiijkhkh
khkhppCC
1
2
)()(
)()(
2
1),(
pi pj
![Page 15: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/15.jpg)
Matching with Shape Contexts
Bipartite Graph Matching
i
ii qpCH ),()( )(
)( 3NOSolved in about
Belongie, et al, PAMI 2002, Shape matching and object recognition using shape contexts
![Page 16: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/16.jpg)
Matching with Shape Contexts
Modeling Transformations
Belongie, et al, PAMI 2002, Shape matching and object recognition using shape contexts
![Page 17: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/17.jpg)
Matching with Shape Contexts
Belongie, et al, PAMI 2002, Shape matching and object recognition using shape contexts
Thin Plane Spline (TPS) Model
(2D Generalization of Cubic Spline)
![Page 18: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/18.jpg)
Matching with Shape ContextsThin Plane Spline (TPS) Model
(2D Generalization of Cubic Spline)
![Page 19: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/19.jpg)
Matching with Shape Contexts
Belongie, et al, PAMI 2002, Shape matching and object recognition using shape contexts
![Page 20: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/20.jpg)
Matching with Shape Contexts
Invariance and Robustness
Belongie, et al, PAMI 2002, Shape matching and object recognition using shape contexts
• Invariant under translation and scaling
• Insensitive to small affine distortion
• Can be made invariant to rotation
![Page 21: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/21.jpg)
Matching with Shape Contexts
Evaluation and Results
Model Point Sets
Deformation Noise Outliers
Target Point Sets
![Page 22: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/22.jpg)
Matching with Shape Contexts
Evaluation and Results
Deformation Noise Outliers
Belongie et al. Chui and Rangarajan Iterated closed point*
![Page 23: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/23.jpg)
Matching with Shape Contexts
Evaluation and Results
![Page 24: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/24.jpg)
Conclusion
![Page 25: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/25.jpg)
Questions
![Page 26: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/26.jpg)
Shape and Image Matching
Shengnan Wang
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 27: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/27.jpg)
Today
• The Pyramid Match Kernel: Discriminative Classification with Sets of Image Features
- Kristen Grauman & Trevor Darrell
- MIT
• Matching Local Self-Similarities across Images and Videos
- Eli Shechtman & Michal Irani
- @ CVPR07
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 28: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/28.jpg)
Set Representation
invariant region
descriptors
local shape features examples under varying
conditions
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 29: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/29.jpg)
Motivation
• How to build a discriminative classifier using the set representation?
• Kernel-based methods (e.g. SVM) are appealing for efficiency and generalization power…
• What determines the appropriates of a kernel?
- Each instance is unordered set of vectors
- Varying number of vectors per instance
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 30: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/30.jpg)
Pyramid Match Kernel
optimal partial
matching
![Page 31: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/31.jpg)
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 32: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/32.jpg)
Example pyramid match
Level 0
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 33: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/33.jpg)
Example pyramid match
Level 1
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 34: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/34.jpg)
Example pyramid match
Level 2
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 35: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/35.jpg)
Pyramid match kernel
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 36: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/36.jpg)
Example pyramid match
pyramid match
optimal match
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 37: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/37.jpg)
Summary: Pyramid match kernel
• linear time complexity: m features of dimension d, L-level pyramid
• model-free
• insensitive to clutter
• positive-definite function
• no independence assumption
• fast, effective object recognition
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 38: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/38.jpg)
Object recognition results
• ETH-80 database :8 object classes
• Features:
– Harris detector
– PCA-SIFT descriptor, d=10
Kernel Complexity Recognition rate
Match [Wallraven et al.]
84%
Bhattacharyya affinity [Kondor &
Jebara]
85%
Pyramid match 84%11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 39: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/39.jpg)
Object recognition results
• Caltech objects database 101 object classes
• Features:– SIFT detector– PCA-SIFT descriptor, d=10
• 30 training images / class• 43% recognition rate
(1% chance performance)• 0.002 seconds per match
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 40: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/40.jpg)
Localization
• Inspect intersections to obtain correspondences between features
• Higher confidence correspondences at finer resolution levels
observationtarget
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 41: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/41.jpg)
Future work
• Geometric constraints
• Fast search of large databases with the pyramid match for image retrieval
• Use as a filter for a slower, explicit correspondence method
• Alternative feature types and classification domains
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 42: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/42.jpg)
Next
• Matching Local Self-Similarities across Images and Videos
- Eli Shechtman & Michal Irani
- @ CVPR07
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 43: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/43.jpg)
What do they do?
How to measure similarity between visual entities (images or videos)
![Page 44: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/44.jpg)
What’s new?
.VS.
• Traditional idea: they share some common visual properties (colors, intensity, gradients, edges or other filter response)
• New idea: they share the same local geometry layout.
Local self repeat
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 45: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/45.jpg)
Self-similarity descriptor
Patch Local Similarity Map Self-similarity descriptor
SSD LPC+Max
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 46: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/46.jpg)
Corresponding Self-similarity
descriptor
G
F
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 47: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/47.jpg)
Self-similarity descriptor
• Benefits
– self-similarity descriptor: local descriptors, wider applicability
– log-polar: accounts for local affine deformations
– maximal correlation value: insensitive to the exact best match position*
– use of patches: more meaningful image patterns
*T. Wolf, L. Bileschi, S. Riesenhuber, M. Poggio, T., Robust Object Recognition with Cortex-Like Mechanisms Serre, IEEE Transactions on Pattern Analysis and Machine Intelligence, 2007
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 48: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/48.jpg)
Matching ensemble of patches*
*O. Boiman and M. Irani. Detecting irregularities in images and in video. In
IEEE International Conference on Computer Vision, Beijing, October 2005.
Center
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 49: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/49.jpg)
Matching ensemble of patches
• Two Implementation Details
– Filter out non-informative descriptors
• Descriptors that do not capture any local self-similarity (salient points)
• Descriptors that contain high self-similarity everywhere (homogeneous region)
– Scale space representation
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 50: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/50.jpg)
Overview
Matching
1. Self-similarity
descriptor
2. Matching ensemble of patches
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 51: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/51.jpg)
Self-similarity descriptor
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 52: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/52.jpg)
Experiments
• Sketch Detection
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 53: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/53.jpg)
Experiments (1)
• Object Detection
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 54: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/54.jpg)
Experiments (2)• Image
Retrieval
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 55: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/55.jpg)
Experiments (3)
• Action detection
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 56: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/56.jpg)
Experiments (4)
• Action detection
11/4/07 Shengnan Wang University of Wisconsin-Madison
![Page 57: Shape Matching - University of Wisconsin–Madisonpages.cs.wisc.edu/~lizhang/courses/cs766-2007f/...–self-similarity descriptor: local descriptors, wider applicability –log-polar:](https://reader036.fdocuments.net/reader036/viewer/2022081607/5eb6f8cc4343525eea72605d/html5/thumbnails/57.jpg)
Questions?
Thank you!
11/4/07 Shengnan Wang University of Wisconsin-Madison