1 / 35

Interest points

Interest points. CSE P 576 Larry Zitnick ( larryz@microsoft.com ) Many slides courtesy of Steve Seitz. How can we find corresponding points?. Not always easy. NASA Mars Rover images. Answer below (look for tiny colored squares…). NASA Mars Rover images

lamar
Download Presentation

Interest points

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Interest points CSE P 576 Larry Zitnick (larryz@microsoft.com) Many slides courtesy of Steve Seitz

  2. How can we find corresponding points?

  3. Not always easy NASA Mars Rover images

  4. Answer below (look for tiny colored squares…) NASA Mars Rover images with SIFT feature matchesFigure by Noah Snavely

  5. Where do humans fixate? Called “fixation points”, and a “saccade” is the process of moving between fixation points. Slowly trace the outline of the above object.

  6. Where do humans fixate? Called “fixation points”, and a “saccade” is the process of moving between fixation points. Slowly trace the outline of the above object.

  7. Where do humans fixate? Result of one subject “me”.

  8. Where do humans fixate? Top down or bottom up? "Eye Movements and Vision" by A. L. Yarbus; Plenum Press, New York; 1967

  9. Want uniqueness Look for image regions that are unusual • Lead to unambiguous matches in other images How to define “unusual”?

  10. Local measures of uniqueness Suppose we only consider a small window of pixels • What defines whether a feature is a good or bad candidate? Slide adapted from Darya Frolova, Denis Simakov, Weizmann Institute.

  11. Feature detection Local measure of feature uniqueness • How does the window change when you shift by a small amount? “corner”:significant change in all directions “flat” region:no change in all directions “edge”: no change along the edge direction Slide adapted from Darya Frolova, Denis Simakov, Weizmann Institute.

  12. Let’s look at the gradient distributions

  13. Principle Component Analysis How to compute PCA components: Subtract off the mean for each data point. Compute the covariance matrix. Compute eigenvectors and eigenvalues. The components are the eigenvectors ranked by the eigenvalues. Principal component is the direction of highest variance. Next, highest component is the direction with highest variance orthogonal to the previous components. Both eigenvalues are large!

  14. Simple example Detect peaks and threshold to find corners.

  15. The math To compute the eigenvalues: Compute the covariance matrix. Compute eigenvalues. Typically Gaussian weights

  16. The Harris operator • - is a variant of the “Harris operator” for feature detection • Very similar to - but less expensive (no square root) • Called the “Harris Corner Detector” or “Harris Operator” • Lots of other detectors, this is one of the most popular Actually used in original paper:

  17. The Harris operator Harris operator

  18. Harris detector example

  19. f value (red high, blue low)

  20. Threshold (f > value)

  21. Find local maxima of f

  22. Harris features (in red)

  23. How can we find correspondences?

  24. How can we find correspondences? Similarity transform

  25. Scale and rotation? Let’s look at scale first: What is the “best” scale?

  26. Scale The Laplacian has a maximal response when the difference between the “inside” and “outside” of the filter is greatest.

  27. Scale Why Gaussian? It is invariant to scale change, i.e., and has several other nice properties. Lindeberg, 1994 In practice, the Laplacian is approximated using a Difference of Gaussian (DoG).

  28. DoG example σ = 1 σ = 66

  29. Scale In practice the image is downsampled for larger sigmas. Lowe, 2004.

  30. How to compute the rotation? Rotation Create edge orientation histogram and find peak.

  31. Harris Laplace Other interest point detectors Scale & Affine Invariant Interest Point Detectors K. MIKOLAJCZYK and C. SCHMID, IJCV 2004

  32. Affine invariant Other interest point detectors Scale & Affine Invariant Interest Point Detectors K. MIKOLAJCZYK and C. SCHMID, IJCV 2004

  33. Computationally efficient Approximate Gaussian filters using box filters: SURF: Speeded Up Robust Features Herbert Bay, TinneTuytelaars, and Luc Van Gool, ECCV 2006

  34. Corner detection by sampling pixels based on decision tree Computationally efficient Machine learning for high-speed corner detection Edward Rosten and Tom Drummond, ECCV 2006

  35. Let’s go to the videos… How well do they work in practice?

More Related