Single view metrology Single view metrology Feature matching vs. tracking Comparing image regions Comparing image regions

Single view metrology Single view metrology Feature matching vs. tracking Comparing image regions Comparing image regions www.phwiki.com

Single view metrology Single view metrology Feature matching vs. tracking Comparing image regions Comparing image regions

Hess, Bill, Military Reporter has reference to this Academic Journal, PHwiki organized this Journal Feature tracking Class 5 Read Section 4.1 of course notes http://www.cs.unc.edu/~marc/tutorial/node49.html Read Shi in addition to Tomasi’s paper on good features to track http://www.unc.edu/courses/2004fall/comp/290/089/papers/shi-tomasi-good-features-cvpr1994.pdf Read Lowe’s paper on SIFT features http://www.unc.edu/courses/2004fall/comp/290/089/papers/Lowe-ijcv03.pdf Don’t as long as get: Assignment 1 (due by next Tuesday be as long as e class) Find a camera Calibration approach 1 Build/use calibration grid (2 orthogonal planes) Per as long as m calibration using (a) DLT in addition to (b) complete gold st in addition to ard algorithm (assume error only in images, model radial distortion, ok to click points by h in addition to ) Calibration approach 2 Build/use planar calibration pattern Use Bouguet’s matlab calibration toolbox (Zhang’s approach) http://www.vision.caltech.edu/bouguetj/calib-doc/ (or implement it yourself as long as extra points) Compare results of approach 1(a),1(b) in addition to 2 Make short report of findings in addition to be ready to discuss in class Single view metrology Allows to relate height of point to height of camera

Erskine College US www.phwiki.com

This Particular University is Related to this Particular Journal

Single view metrology Allows to transfer point from one plane to another Feature tracking Class 5 Read Section 4.1 of course notes http://www.cs.unc.edu/~marc/tutorial/node49.html Read Shi in addition to Tomasi’s paper on good features to track http://www.unc.edu/courses/2004fall/comp/290/089/papers/shi-tomasi-good-features-cvpr1994.pdf Read Lowe’s paper on SIFT features http://www.unc.edu/courses/2004fall/comp/290/089/papers/Lowe-ijcv03.pdf Feature matching vs. tracking Extract features independently in addition to then match by comparing descriptors Extract features in first images in addition to then try to find same feature back in next view What is a good feature Image-to-image correspondences are key to passive triangulation-based 3D reconstruction

Comparing image regions Compare intensities pixel-by-pixel I(x,y) I´(x,y) Sum of Square Differences Dissimilarity measures Comparing image regions Compare intensities pixel-by-pixel I(x,y) I´(x,y) Zero-mean Normalized Cross Correlation Similarity measures Feature points Required properties: Well-defined (i.e. neigboring points should all be different) Stable across views (i.e. same 3D point should be extracted as feature as long as neighboring viewpoints)

Feature point extraction homogeneous edge corner Find points that differ as much as possible from all neighboring points Feature point extraction Approximate SSD as long as small displacement Image difference, square difference as long as pixel SSD as long as window Feature point extraction homogeneous edge corner Find points as long as which the following is maximum i.e. maximize smallest eigenvalue of M

Harris corner detector Only use local maxima, subpixel accuracy through second order surface fitting Select strongest features over whole image in addition to over each tile (e.g. 1000/image, 2/tile) Use small local window: Maximize „cornerness“: Simple matching as long as each corner in image 1 find the corner in image 2 that is most similar (using SSD or NCC) in addition to vice-versa Only compare geometrically compatible points Keep mutual best matches What trans as long as mations does this work as long as Feature matching: example What trans as long as mations does this work as long as What level of trans as long as mation do we need

Wide baseline matching Requirement to cope with larger variations between images Translation, rotation, scaling Foreshortening Non-diffuse reflections Illumination geometric trans as long as mations photometric changes Wide-baseline matching example (Tuytelaars in addition to Van Gool BMVC 2000) Lowe’s SIFT features Recover features with position, orientation in addition to scale (Lowe, ICCV99)

Position Look as long as strong responses of DOG filter (Difference-Of-Gaussian) Only consider local maxima Scale Look as long as strong responses of DOG filter (Difference-Of-Gaussian) over scale space Only consider local maxima in both position in addition to scale Fit quadratic around maxima as long as subpixel Orientation Create histogram of local gradient directions computed at selected scale Assign canonical orientation at peak of smoothed histogram Each key specifies stable 2D coordinates (x, y, scale, orientation)

Minimum contrast in addition to “cornerness” SIFT descriptor Thresholded image gradients are sampled over 16×16 array of locations in scale space Create array of orientation histograms 8 orientations x 4×4 histogram array = 128 dimensions

Hess, Bill Sierra Vista Herald Military Reporter www.phwiki.com

Matas et al.’s maximally stable regions Look as long as extremal regions http://cmp.felk.cvut.cz/~matas/papers/matas-bmvc02.pdf Mikolaczyk in addition to Schmid LoG Features Feature tracking Identify features in addition to track them over video Small difference between frames potential large difference overall St in addition to ard approach: KLT (Kanade-Lukas-Tomasi)

Good features to track Use same window in feature selection as as long as tracking itself Compute motion assuming it is small Affine is also possible, but a bit harder (6×6 in stead of 2×2) Example Example

Next class: triangulation in addition to reconstruction C1 m1 L1 Triangulation calibration correspondences

Hess, Bill Military Reporter

Hess, Bill is from United States and they belong to Sierra Vista Herald and they are from  Sierra Vista, United States got related to this Particular Journal. and Hess, Bill deal with the subjects like Military and Armed Forces

Journal Ratings by Erskine College

This Particular Journal got reviewed and rated by Erskine College and short form of this particular Institution is US and gave this Journal an Excellent Rating.