cse 185 introduction to computer vision feature matching
TRANSCRIPT
![Page 1: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/1.jpg)
CSE 185 Introduction to Computer
VisionFeature Matching
![Page 2: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/2.jpg)
Feature matching
• Correspondence: matching points, patches, edges, or regions across images
≈
![Page 3: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/3.jpg)
Keypoint matching
AfBf
B1
B2
B3A1
A2 A3
Tffd BA ),(
1. Find a set of distinctive key- points
3. Extract and normalize the region content
2. Define a region around each keypoint
4. Compute a local descriptor from the normalized region5. Match local descriptors
![Page 4: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/4.jpg)
Review: Interest points
• Keypoint detection: repeatable and distinctive– Corners, blobs, stable
regions– Harris, DoG, MSER– SIFT
![Page 5: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/5.jpg)
Which interest detector• What do you want it for?
– Precise localization in x-y: Harris– Good localization in scale: Difference of Gaussian– Flexible region shape: MSER
• Best choice often application dependent– Harris-/Hessian-Laplace/DoG work well for many natural
categories– MSER works well for buildings and printed things
• Why choose?– Get more points with more detectors
• There have been extensive evaluations/comparisons– [Mikolajczyk et al., IJCV’05, PAMI’05]– All detectors/descriptors shown here work well
![Page 6: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/6.jpg)
Local feature descriptors
• Most features can be thought of as templates, histograms (counts), or combinations
• The ideal descriptor should be– Robust and distinctive– Compact and efficient
• Most available descriptors focus on edge/gradient information– Capture texture information– Color rarely used
![Page 7: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/7.jpg)
How do we decide which features match?
![Page 8: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/8.jpg)
Feature matching
• Szeliski 4.1.3– Simple feature-space methods– Evaluation methods– Acceleration methods– Geometric verification (Chapter 6)
![Page 9: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/9.jpg)
Feature matching
• Simple criteria: One feature matches to another if those features are nearest neighbors and their distance is below some threshold.
• Problems:– Threshold is difficult to set– Non-distinctive features could have lots
of close matches, only one of which is correct
![Page 10: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/10.jpg)
Matching local features
• Threshold based on the ratio of 1st nearest neighbor to 2nd nearest neighbor distance
Reject all matches in whichthe distance ration > 0.8, which eliminates 90% of false matcheswhile discarding less than 5% correct matches
![Page 11: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/11.jpg)
SIFT repeatability
It shows the stability of detection for keypoint location, orientation, and final matching to a database as a function of affine distortion. The degree of affine distortion is expressedin terms of the equivalent viewpoint rotation in depth for a planar surface.
![Page 12: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/12.jpg)
Fitting: find the parameters of a model that best fit the data
Alignment: find the parameters of the transformation that best align matched points
Fitting and alignment
![Page 13: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/13.jpg)
Checkerboard
• Often used in camera calibration
![Page 14: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/14.jpg)
Fitting and alignment
• Design challenges– Design a suitable goodness of fit
measure• Similarity should reflect application goals• Encode robustness to outliers and noise
– Design an optimization method• Avoid local optima• Find best parameters quickly
![Page 15: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/15.jpg)
Fitting and alignment: Methods
• Global optimization / Search for parameters– Least squares fit– Robust least squares– Iterative closest point (ICP)
• Hypothesize and test– Generalized Hough transform– RANSAC
![Page 16: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/16.jpg)
Least squares line fitting• Data: (x1, y1), …, (xn, yn)
• Line equation: yi = m xi + b
• Find (m, b) to minimize
022 yAApA TT
dp
dE
n
i ii bxmyE1
2)(
(xi, yi)
y=mx+b
yAAApyAApA TTTT 1
Matlab: p = A \ y;
![Page 17: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/17.jpg)
Least squares (global) optimizationGood• Clearly specified objective• Optimization is easy
Bad• May not be what you want to optimize • Sensitive to outliers
– Bad matches, extra points
• Doesn’t allow you to get multiple good fits– Detecting multiple objects, lines, etc.
![Page 18: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/18.jpg)
Hypothesize and test1. Propose parameters
– Try all possible– Each point votes for all consistent parameters– Repeatedly sample enough points to solve for
parameters
2. Score the given parameters– Number of consistent points, possibly weighted by
distance
3. Choose from among the set of parameters– Global or local maximum of scores
4. Possibly refine parameters using inliers
![Page 19: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/19.jpg)
Hough transform: Outline
1. Create a grid of parameter values
2. Each point votes for a set of parameters, incrementing those values in grid
3. Find maximum or local maxima in grid
![Page 20: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/20.jpg)
x
y
b
m
y = m x + b
Given a set of points, find the curve or line that explains the data points best
Hough space
Hough transform
Duality: Each point has a dual line in the parameter space
m = -(1/x)b + y/x
![Page 21: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/21.jpg)
x
y
b
m
x
y m3 5 3 3 2 2
3 7 11 10 4 3
2 3 1 4 5 2
2 1 0 1 3 3
b
Hough transform
![Page 22: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/22.jpg)
x
y
Issue : parameter space [m,b] is unbounded…
Hough space
)sin
()sin
cos(
r
xy
Use a polar representation for the parameter space
Hough transform
Duality: Each point has a dual curve in the parameter space
sincos 00 yxr
![Page 23: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/23.jpg)
features votes
Hough transform: Experiments
![Page 24: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/24.jpg)
Hough transform
![Page 25: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/25.jpg)
features votes
Need to adjust grid size or smooth
Noisy data
Hough transform: Experiments
![Page 26: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/26.jpg)
Issue: spurious peaks due to uniform noise
features votes
Hough transform: Experiments
![Page 27: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/27.jpg)
1. Image Canny
![Page 28: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/28.jpg)
2. Canny Hough votes
![Page 29: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/29.jpg)
3. Hough votes Edges
Find peaks and post-process
![Page 30: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/30.jpg)
Hough transform example
![Page 31: CSE 185 Introduction to Computer Vision Feature Matching](https://reader036.vdocuments.net/reader036/viewer/2022062305/5697c0151a28abf838ccdef5/html5/thumbnails/31.jpg)
Hough transform: Finding lines• Using m,b parameterization• Using r, theta parameterization
– Using oriented gradients
• Practical considerations– Bin size– Smoothing– Finding multiple lines– Finding line segments