We think you have liked this presentation. If you wish to download it, please recommend it to your friends in any social system. Share buttons are a little bit lower. Thank you!
Presentation is loading. Please wait.
Published byHarry Douglas
Modified about 1 year ago
81.3% Affine invariant feature detector and image correlation Martin Bujňák © 2004 Martin Bujňák,
Previous work… (simple and fast) Moravec detector –using small window –determining the average changes of image intensity –big-change implies feature © 2004 Martin Bujňák,
Previous work… (simple and fast) Harris, Stephens –analytically described and enhanced Moravec detector –correlation with neighbors replaced with simple calculations - fast © 2004 Martin Bujňák,
Works fine but,… good results, but –region is described by one point –additional information about local shape can be obtained by principal component analysis would help, but only for small very changes even though usable but –many feature points must be detected and robust brute force algorithm for image correspondence must be used (RANSAC…) © 2004 Martin Bujňák,
Idea Let the feature point be rotation, scale and translation invariant (or other similarity transformations)… –not the miracle, but we get more stable features –let the feature by intersection of "bigger or stronger" edge (figure 1) figure 1. Features in edge intersection. © 2004 Martin Bujňák,
Idea Build graph –feature is graph node –two features are connected if there exists direct (or long A-B are connected) edge between them. Edges are not oriented. A B X1 X2 X3 © 2004 Martin Bujňák,
Idea Feature information –neighborhood with other vertices – ordered by angle –distance ration to neighbor for each connection (marked red) A n1n2 n3 φ2φ2 φ1φ1 φ3φ3 φ4φ4 Φ = 0 © 2004 Martin Bujňák,
Advanced – pie – slice correlation get elliptical cut bounded by two edges and elliptical arc (and call it pie slice) A n1 n3 φ2φ2 φ1φ1 φ3φ3 φ4φ4 Φ = 0 n3-n1 pie slice Feature © 2004 Martin Bujňák,
Advanced – pie – slice correlation Correlation process –two features are at the first rotated to best fit incidence information –correlated each pie-slice separately (using standard similarity or dissimilarity measures) © 2004 Martin Bujňák,
Advanced – pie – slice correlation Results –more stable for affine and also for projective transformation –good “corners” will remain –feature don’t need to well correlate in all slices connectivity information helps graph topology speeds up RANSAC elimination pie-slice correlation give better matching probability than standard one. © 2004 Martin Bujňák,
H.P. Moravec, Visual mapping by a robot rover, Proc. of the 6th International Joint Conference on Artificial Intelligence, pp , 1979 C. Harris - M. Stephens, A combined corner and edge detector, Fourth Alvey Vision Conference, pp , 1988 R. Deriche - G. Giraudon, "A computational approach for corner and vertex detection", International Journal of Computer Vision, 1(2): , 1993 S.M. Smith - J.M. Brady. "SUSAN - a new approach to low level image processing". Int. Journal of Computer Vision, Vol.23, Nr.1, pp.45-78, 1997 K. Mikolajczyk - S.M. Smith. "Scale & affine invariant interest point detectors". INRIA Rhne-Alpes GRAVIR-CNRS, 655 av. de l’Europe, Montbonnot, France T. Kadir - A. Zisserman - M.Brady. "An affinne invariant salient region detector". Department of Engineering Science, University of Oxford, UK. M. Pollefeys - L. Van Gool - M. Vergauwen - F. Verbiest - K. Cornelis - J. Tops - R. Koch, Visual modeling with a hand-held camera, International Journal of Computer Vision 59(3), , 2004 References © 2004 Martin Bujňák,
Computer Vision : CISC 4/689 Corner Detection Basic idea: Find points where two edges meeti.e., high gradient in two directions Cornerness is undefined.
Feature extraction: Corners 9300 Harris Corners Pkwy, Charlotte, NC.
Feature points extraction Many slides are courtesy of Darya Frolova, Denis Simakov A low level building block in many applications: Structure from motion.
Feature Detection. Description Localization More Points Robust to occlusion Works with less texture More Repeatable Robust detection Precise localization.
Scale & Affine Invariant Interest Point Detectors Mikolajczyk & Schmid presented by Dustin Lennon.
Object Recognition from Local Scale-Invariant Features David G. Lowe Presented by Ashley L. Kapron.
Distinctive Image Features from Scale-Invariant Keypoints.
Feature Based Image Mosaicing Satya Prakash Mallick.
Lecture 13 Image features Antonio Torralba, 2013 With some slides from Darya Frolova, Denis Simakov, David Lowe, Bill Freeman.
CSCE 643 Computer Vision: Lucas-Kanade Registration Jinxiang Chai.
Institut für Elektrische Meßtechnik und Meßsignalverarbeitung Professor Horst Cerjak, Augmented Reality VU 3 Algorithms Axel Pinz.
Feature matching “What stuff in the left image matches with stuff on the right?” Necessary for automatic panorama stitching (Part of Project 4!) Slides.
Distinctive Image Features from Scale-Invariant Keypoints David Lowe.
Object Recognition Using Locality-Sensitive Hashing of Shape Contexts Andrea Frome, Jitendra Malik Presented by Ilias Apostolopoulos.
Optical Character Recognition for Handwritten Characters Giorgos Vamvakas National Center for Scientific Research “Demokritos” Athens - Greece Institute.
CS 336 March 19, 2012 Tandy Warnow. Basic Graph Terminology Nodes, vertices, edges, degrees, paths, cycles, connected components, adjacency, isolated.
Dilations. Dilation Scale Factor Center of Dilation.
0 By: Navid Einackchi Isfahan University of Technology Electrical and Computer Department Spring 2007 Image alignment and stitching using Object Recognition.
Ter Haar Romeny, ICPR 2010 Introduction to Scale-Space and Deep Structure.
Modeling Modeling is simply the process of creating 3D objects –Many different processes to create models –Many different representations of model data.
Surface normals and principal component analysis (PCA) 3DM slides by Marc van Kreveld 1.
Coherent Laplacian 3D protrusion segmentation Oxford Brookes Vision Group Queen Mary, University of London, 11/12/2009 Fabio Cuzzolin.
Goal: a graph representation of the topology of a gray scale image. The graph represents the hierarchy of the lower and upper level sets of the gray level.
Face Recognition Sumitha Balasuriya. Computer Vision Image processing is a precursor to Computer Vision – making a computer understand and interpret whats.
Computer vision: models, learning and inference Chapter 15 Models for transformations.
Drafting – Product Design & Architecture Orthographic Projection.
Image Registration Mapping of Evolution. Registration Goals Assume the correspondences are known Find such f() and g() such that the images are best.
Epipolar Geometry. Two-view geometry Epipolar geometry 3D reconstruction F-matrix comp. Structure comp.
Richard Young Richard Young Optronic Laboratories Kathleen Muray Kathleen Muray INPHORA Carolyn Jones Carolyn Jones CJ Enterprises.
Course Overview What is AI? What are the Major Challenges? What are the Main Techniques? Where are we failing, and why? Step back and look at.
© 2016 SlidePlayer.com Inc. All rights reserved.