# 102215 Interest Points Galatea of the Spheres Salvador

- Slides: 55

10/22/15 Interest Points Galatea of the Spheres Salvador Dali Computational Photography Derek Hoiem, University of Illinois

Today’s class • Review of “Modeling the Physical World” • Interest points

Pinhole camera model • Linear projection from 3 D to 2 D – Be familiar with projection matrix (focal length, principal point, etc. ) . Optical Center (u 0, v 0) . . v u f . Camera Center (tx, ty, tz) Z Y

Vanishing points and metrology • Parallel lines in 3 D intersect at a vanishing point in the image Vertical vanishing point (at infinity) Vanishing line Vanishing point • Can measure relative object heights using vanishing point tricks 5 4 3 2 1

Single-view 3 D Reconstruction • Technically impossible to go from 2 D to 3 D, but we can do it with simplifying models – Need some interaction or recognition algorithms – Uses basic VP tricks and projective geometry

Lens, aperture, focal length • Aperture size and focal length control amount of exposure needed, depth of field, field of view Good explanation: http: //www. cambridgeincolour. com/tutorials/depth-of-field. htm

Capturing light with a mirrored sphere

One small snag • How do we deal with light sources? Sun, lights, etc? – They are much, much brighter than the rest of the environment . 46 1907 Relative Brightness . 15116 1 . . 18 . • Use High Dynamic Range photography

Key ideas for Image-based Lighting • Capturing HDR images: needed so that light probes capture full range of radiance

Key ideas for Image-based Lighting • Relighting: environment map acts as light source, substituting for distant scene

Next section of topics • Correspondence – How do we find matching patches in two images? – How can we automatically align two images of the same scene? – How do we find images with similar content? – How do we tell if two pictures are of the same person’s face? – How can we detect objects from a particular category? • Applications – – Photo stitching Object recognition 3 D Reconstruction Tracking

How can we align two pictures? • Case of global transformation

How can we align two pictures? • Global matching? – But what if • Not just translation change, but rotation and scale? • Only small pieces of the pictures match?

Today: Keypoint Matching 1. Find a set of distinctive keypoints B 3 A 1 A 2 2. Define a region around each keypoint A 3 B 2 B 1 3. Extract and normalize the region content 4. Compute a local descriptor from the normalized region K. Grauman, B. Leibe 5. Match local descriptors

Main challenges • Change in position, scale, and rotation • Change in viewpoint • Occlusion • Articulation, change in appearance

Question • Why not just take every patch in the original image and find best match in second image?

Goals for Keypoints Detect points that are repeatable and distinctive

Key trade-offs A B 1 3 A A 2 3 B Localization B 2 1 More Points More Repeatable Robust to occlusion Works with less texture Robust detection Precise localization Description More Robust More Selective Deal with expected variations Maximize correct matches Minimize wrong matches

Keypoint localization • Suppose you have to click on some point, go away and come back after I deform the image, and click on the same points again. original – Which points would you choose? deformed

Keypoint localization • Goals: – Repeatable detection – Precise localization – Interesting content K. Grauman, B. Leibe

Choosing interest points Where would you tell your friend to meet you?

Choosing interest points Where would you tell your friend to meet you?

Choosing interest points • Corners • Peaks/Valleys

Which patches are easier to match? ?

Choosing interest points • If you wanted to meet a friend would you say a) “Let’s meet on campus. ” b) “Let’s meet on Green street. ” c) “Let’s meet at Green and Wright. ” • Or if you were in a secluded area: a) “Let’s meet in the Plains of Akbar. ” b) “Let’s meet on the side of Mt. Doom. ” c) “Let’s meet on top of Mt. Doom. ”

Many Existing Detectors Available Hessian & Harris [Beaudet ‘ 78], [Harris ‘ 88] Laplacian, Do. G [Lindeberg ‘ 98], [Lowe 1999] Harris-/Hessian-Laplace [Mikolajczyk & Schmid ‘ 01] Harris-/Hessian-Affine[Mikolajczyk & Schmid ‘ 04] EBR and IBR [Tuytelaars & Van Gool ‘ 04] MSER [Matas ‘ 02] Salient Regions [Kadir & Brady ‘ 01] Others… K. Grauman, B. Leibe

Harris Detector [Harris 88] Second moment matrix Intuition: Search for local neighborhoods where the image gradient has two main directions (eigenvectors). K. Grauman, B. Leibe

Harris Detector [Harris 88] Second moment matrix Ix Iy Ix 2 Iy 2 Ix Iy g(Ix 2) g(Iy 2) g(Ix. Iy) 1. Image derivatives 2. Square of derivatives 3. Gaussian filter g(σI) 4. Cornerness function – both eigenvalues are strong 5. Non-maxima suppression g(Ix. Iy) 28 har

Matlab code for Harris Detector function [ptx, pty] = detect. Keypoints(im, alpha, N) % get harris function gfil = fspecial('gaussian', [7 7], 1); % smoothing filter imblur = imfilter(im, gfil); % smooth image [Ix, Iy] = gradient(imblur); % compute gradient Ixx = imfilter(Ix. *Ix, gfil); % compute smoothed x-gradient sq Iyy = imfilter(Iy. *Iy, gfil); % compute smoothed y-gradient sq Ixy = imfilter(Ix. *Iy, gfil); har = Ixx. *Iyy - Ixy. *Ixy - alpha*(Ixx+Iyy). ^2; % cornerness % get local maxima within 7 x 7 window maxv = ordfilt 2(har, 49, ones(7)); % sorts values in each window maxv 2 = ordfilt 2(har, 48, ones(7)); ind = find(maxv==har & maxv~=maxv 2); % get top N points [sv, sind] = sort(har(ind), 'descend'); sind = ind(sind); [pty, ptx] = ind 2 sub(size(im), sind(1: min(N, numel(sind))));

Harris Detector – Responses [Harris 88] Effect: A very precise corner detector.

Harris Detector – Responses [Harris 88]

So far: can localize in x-y, but not scale

Automatic Scale Selection How to find corresponding patch sizes? K. Grauman, B. Leibe

Automatic Scale Selection • Function responses for increasing scale (scale signature) K. Grauman, B. Leibe

Automatic Scale Selection • Function responses for increasing scale (scale signature) K. Grauman, B. Leibe

Automatic Scale Selection • Function responses for increasing scale (scale signature) K. Grauman, B. Leibe

What Is A Useful Signature Function? • Difference of Gaussian = “blob” detector K. Grauman, B. Leibe

Difference-of-Gaussian (Do. G) = - K. Grauman, B. Leibe

Do. G – Efficient Computation • Computation in Gaussian scale pyramid Sampling with step σ4 =2 σ σ σ Original image σ K. Grauman, B. Leibe

Results: Lowe’s Do. G K. Grauman, B. Leibe

Orientation Normalization • Compute orientation histogram [Lowe, SIFT, 1999] • Select dominant orientation • Normalize: rotate to fixed orientation 0 T. Tuytelaars, B. Leibe 2π

Available at a web site near you… • For most local feature detectors, executables are available online: – http: //robots. ox. ac. uk/~vgg/research/affine – http: //www. cs. ubc. ca/~lowe/keypoints/ – http: //www. vision. ee. ethz. ch/~surf K. Grauman, B. Leibe

How do we describe the keypoint?

Descriptors for local matching • Image patch (plain intensities or gradientbased features) Example of patch-based matching for stereo

Local descriptors for matching different views/times • The ideal descriptor should be – Robust to expected deformation – Distinctive – Compact – Efficient to compute • Most available descriptors focus on edge/gradient information – Capture texture information – Color rarely used K. Grauman, B. Leibe

Local Descriptors: SIFT Descriptor [Lowe, ICCV 1999] Histogram of oriented gradients • Captures important texture information • Robust to small translations / affine deformations K. Grauman, B. Leibe

Details of Lowe’s SIFT algorithm • Run Do. G detector – Find maxima in location/scale space – Remove edge points • Find all major orientations – Bin orientations into 36 bin histogram • Weight by gradient magnitude • Weight by distance to center (Gaussian-weighted mean) – Return orientations within 0. 8 of peak • Use parabola for better orientation fit • For each (x, y, scale, orientation), create descriptor: – – Sample 16 x 16 gradient mag. and rel. orientation Bin 4 x 4 samples into 4 x 4 histograms Threshold values to max of 0. 2, divide by L 2 norm Final descriptor: 4 x 4 x 8 normalized histograms Lowe IJCV 2004

Matching SIFT Descriptors • Nearest neighbor (Euclidean distance) • Threshold ratio of nearest to 2 nd nearest descriptor Lowe IJCV 2004

Local Descriptors: SURF • Fast approximation of SIFT idea ➢ ➢ Efficient computation by 2 D box filters & integral images ⇒ 6 times faster than SIFT Equivalent quality for object identification • GPU implementation available ➢ ➢ [Bay, ECCV’ 06], [Cornelis, CVGPU’ 08] Feature extraction @ 200 Hz (detector + descriptor, 640× 480 img) http: //www. vision. ee. ethz. ch/~surf K. Grauman, B. Leibe

What to use when? Detectors • Harris gives very precise localization but doesn’t predict scale – Good for some tracking applications • DOG (difference of Gaussian) provides ok localization and scale – Good for multi-scale or long-range matching Descriptors • Intensity patch: suitable for precise local search • SIFT: good for long-range matching, general descriptor

Things to remember • Keypoint detection: repeatable and distinctive – Corners, blobs – Harris, Do. G • Descriptors: robust and selective – SIFT: spatial histograms of gradient orientation

Next time: Panoramic Stitching Camera Center