1 / 22

SURF: Speeded Up Robust Features, ECCV 2006. Herbert Bay, Tinne Tuytelaars, and Luc Van Gool

SURF: Speeded Up Robust Features, ECCV 2006. Herbert Bay, Tinne Tuytelaars, and Luc Van Gool. Group Meeting Presented by Wyman 10/14/2006. Background. Local invariant Interest point detector-descriptor For finding correspondences between two images of the same scene or object

dolf
Télécharger la présentation

SURF: Speeded Up Robust Features, ECCV 2006. Herbert Bay, Tinne Tuytelaars, and Luc Van Gool

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. SURF: Speeded Up Robust Features, ECCV 2006.Herbert Bay, Tinne Tuytelaars, and Luc Van Gool Group Meeting Presented by Wyman 10/14/2006

  2. Background • Local invariant Interest point detector-descriptor • For finding correspondences between two images of the same scene or object • Many applications, including 3D reconstruction, image retrieval and object recognition • SIFT is one of the best but slow • Image of size 1000 x 700 described in around 6 seconds (actual cost depends on the # features generated, 4000 in this case) • 128-D feature vectors

  3. Motivation • Fast interest point detection • Distinctive interest point description • Speeded-up descriptor matching • Invariant to common image transformations: • Image rotation • Scale changes • Illumination change • Small change in Viewpoint

  4. Methodology • Using integral images for major speed up • Integral Image (summed area tables) is an intermediate representation for the image and contains the sum of gray scale pixel values of image • Second order derivative and Haar-wavelet response Cost four additions operation only

  5. Detection • Hessian-based interest point localization • Lxx(x,y,σ) is the Laplacian of Gaussian of the image • It is the convolution of the Gaussian second order derivative with the image • Lindeberg showed Gaussian function is optimal for scale-space analysis • This paper argues that Gaussian is overrated since the property that no new structures can appear while going to lower resolution is not proven in 2D case

  6. Detection • Approximated second order derivatives with box filters (mean/average filter)

  7. Detection • Scale analysis with constant image size 9 x 9, 15 x 15, 21 x 21, 27 x 27  39 x 39, 51 x 51 … 1st octave 2nd octave

  8. Detection • Non-maximum suppression and interpolation • Blob-like feature detector

  9. Description • Orientation Assignment Circular neighborhood of radius 6s around the interest point (s = the scale at which the point was detected) x response y response Side length = 4s Cost 6 operation to compute the response

  10. Description • Dominant orientation • The Haar wavelet responses are represented as vectors • Sum all responses withina sliding orientationwindow covering an angle of 60 degree • The two summed response yield a new vector • The longest vector is the dominant orientation • Second longest is … ignored

  11. Description • Split the interest region up into 4 x 4 square sub-regions with 5 x 5 regularly spaced sample points inside • Calculate Haar wavelet response dx and dy • Weight the response with a Gaussian kernel centered at the interest point • Sum the response over each sub-region for dx and dy separately  feature vector of length 32 • In order to bring in information about the polarity of the intensity changes, extract the sum of absolute value of the responses  feature vector of length 64 • Normalize the vector into unit length

  12. Description

  13. Description • SURF-128 • The sum of dx and |dx| are computed separately for dy < 0 and dy >0 • Similarly for the sum of dy and |dy| • This doubles the length of a feature vector

  14. Matching • Fast indexing through the sign of the Laplacian for the underlying interest point • The sign of trace of the Hessian matrix • Trace = Lxx + Lyy • Either 0 or 1 (Hard thresholding, may have boundary effect …) • In the matching stage, compare features if they have the same type of contrast (sign)

  15. Experimental Results

  16. Experimental Results Viewpoint change of 30 degrees

  17. Experimental Results

  18. Experimental Results 1. Wall 2. Boat 3. Bikes 4. Trees

  19. Analysis • I have carried out a benchmark on SURF and SIFT using the Visual Geometry Group Dataset • SURF: Fast-Hessian detector + SURF descriptor • SIFT: DOG detector + SIFT descriptor

  20. Analysis

  21. Analysis • SURF is good at • handling serious blurring • handling image rotation • SURF is poor at • handling viewpoint change • handling illumination change • SURF is always better than the SIFT implemented by VGG but not the original SIFT

  22. Conclusion • SURF describes image faster than SIFT by 3 times • SURF is not as well as SIFT on invariance to illumination change and viewpoint change

More Related