Feature extraction: Corners and blobs

Similar documents
Keypoint extraction: Corners Harris Corners Pkwy, Charlotte, NC

Interest Point Detection. Lecture-4

Feature extraction: Corners and blobs

Edges and Scale. Image Features. Detecting edges. Origin of Edges. Solution: smooth first. Effects of noise

Instance-level recognition: Local invariant features. Cordelia Schmid INRIA, Grenoble

Recap: edge detection. Source: D. Lowe, L. Fei-Fei

CEE598 - Visual Sensing for Civil Infrastructure Eng. & Mgmt.

Corners, Blobs & Descriptors. With slides from S. Lazebnik & S. Seitz, D. Lowe, A. Efros

Image Analysis. Feature extraction: corners and blobs

Lecture 05 Point Feature Detection and Matching

Instance-level recognition: Local invariant features. Cordelia Schmid INRIA, Grenoble

Instance-level l recognition. Cordelia Schmid INRIA

Perception III: Filtering, Edges, and Point-features

Instance-level recognition: Local invariant features. Cordelia Schmid INRIA, Grenoble

Blob Detection CSC 767

SIFT keypoint detection. D. Lowe, Distinctive image features from scale-invariant keypoints, IJCV 60 (2), pp , 2004.

6.869 Advances in Computer Vision. Prof. Bill Freeman March 1, 2005

Lecture 8: Interest Point Detection. Saad J Bedros

Feature detectors and descriptors. Fei-Fei Li

Feature detectors and descriptors. Fei-Fei Li

* h + = Lec 05: Interesting Points Detection. Image Analysis & Retrieval. Outline. Image Filtering. Recap of Lec 04 Image Filtering Edge Features

Detectors part II Descriptors

Blobs & Scale Invariance

Instance-level l recognition. Cordelia Schmid & Josef Sivic INRIA

Advances in Computer Vision. Prof. Bill Freeman. Image and shape descriptors. Readings: Mikolajczyk and Schmid; Belongie et al.

Lecture 8: Interest Point Detection. Saad J Bedros

CS4670: Computer Vision Kavita Bala. Lecture 7: Harris Corner Detec=on

EE2 Mathematics : Functions of Multiple Variables

Optical flow. Subhransu Maji. CMPSCI 670: Computer Vision. October 20, 2016

Local Features (contd.)

Lecture 12. Local Feature Detection. Matching with Invariant Features. Why extract features? Why extract features? Why extract features?

CS5670: Computer Vision

MAT389 Fall 2016, Problem Set 6

CSE 473/573 Computer Vision and Image Processing (CVIP)

Properties of detectors Edge detectors Harris DoG Properties of descriptors SIFT HOG Shape context

Geometric Image Manipulation. Lecture #4 Wednesday, January 24, 2018

LoG Blob Finding and Scale. Scale Selection. Blobs (and scale selection) Achieving scale covariance. Blob detection in 2D. Blob detection in 2D

Invariant local features. Invariant Local Features. Classes of transformations. (Good) invariant local features. Case study: panorama stitching

INTEREST POINTS AT DIFFERENT SCALES

Chapter 6 Momentum Transfer in an External Laminar Boundary Layer

Linear Strain Triangle and other types of 2D elements. By S. Ziaei Rad

Achieving scale covariance

Extract useful building blocks: blobs. the same image like for the corners

Overview. Introduction to local features. Harris interest points + SSD, ZNCC, SIFT. Evaluation and comparison of different detectors

Interest Operators. All lectures are from posted research papers. Harris Corner Detector: the first and most basic interest operator

10.4 Solving Equations in Quadratic Form, Equations Reducible to Quadratics

u P(t) = P(x,y) r v t=0 4/4/2006 Motion ( F.Robilliard) 1

Overview. Harris interest points. Comparing interest points (SSD, ZNCC, SIFT) Scale & affine invariant interest points

Vectors in Rn un. This definition of norm is an extension of the Pythagorean Theorem. Consider the vector u = (5, 8) in R 2

Change of Variables. (f T) JT. f = U

PHASE PLANE DIAGRAMS OF DIFFERENCE EQUATIONS. 1. Introduction

Concept of Stress at a Point

10.2 Solving Quadratic Equations by Completing the Square

Corner detection: the basic idea

1 The space of linear transformations from R n to R m :

Lesson 81: The Cross Product of Vectors

Motion estimation. Digital Visual Effects Yung-Yu Chuang. with slides by Michael Black and P. Anandan

Formal Methods for Deriving Element Equations

Visualisations of Gussian and Mean Curvatures by Using Mathematica and webmathematica

Vlad Estivill-Castro (2016) Robots for People --- A project for intelligent integrated systems

Chapter 1: Differential Form of Basic Equations

Scale & Affine Invariant Interest Point Detectors

SUBJECT:ENGINEERING MATHEMATICS-I SUBJECT CODE :SMT1101 UNIT III FUNCTIONS OF SEVERAL VARIABLES. Jacobians

Algebraic Multigrid. Multigrid

SURF Features. Jacky Baltes Dept. of Computer Science University of Manitoba WWW:

CS 3710: Visual Recognition Describing Images with Features. Adriana Kovashka Department of Computer Science January 8, 2015

3.3 Operations With Vectors, Linear Combinations

Primary dependent variable is fluid velocity vector V = V ( r ); where r is the position vector

Section 7.4: Integration of Rational Functions by Partial Fractions

ON TRANSIENT DYNAMICS, OFF-EQUILIBRIUM BEHAVIOUR AND IDENTIFICATION IN BLENDED MULTIPLE MODEL STRUCTURES

Lecture 6: Finding Features (part 1/2)

Elements of Coordinate System Transformations

We automate the bivariate change-of-variables technique for bivariate continuous random variables with

called the potential flow, and function φ is called the velocity potential.

1 Differential Equations for Solid Mechanics

Reduction of over-determined systems of differential equations

Digital Image Processing. Lecture 8 (Enhancement in the Frequency domain) Bu-Ali Sina University Computer Engineering Dep.

Change of Variables. f(x, y) da = (1) If the transformation T hasn t already been given, come up with the transformation to use.

Image Processing 1 (IP1) Bildverarbeitung 1

Math 263 Assignment #3 Solutions. 1. A function z = f(x, y) is called harmonic if it satisfies Laplace s equation:

CS 450: COMPUTER GRAPHICS VECTORS SPRING 2016 DR. MICHAEL J. REALE

Lecture 7: Finding Features (part 2/2)

Graphs and Networks Lecture 5. PageRank. Lecturer: Daniel A. Spielman September 20, 2007

EXERCISES WAVE EQUATION. In Problems 1 and 2 solve the heat equation (1) subject to the given conditions. Assume a rod of length L.

Chapter 2 Difficulties associated with corners

m = Average Rate of Change (Secant Slope) Example:

Image matching. by Diva Sian. by swashford

Quiz #20. y 2 + 6y = 12x y 2 + 6y + 9 = 12x (y + 3) 2 = 12x + 24 (y + 3) 2 = 12(x 2)

Edge Detection. Image Processing - Computer Vision

Spring, 2008 CIS 610. Advanced Geometric Methods in Computer Science Jean Gallier Homework 1, Corrected Version

Diagonalization of Quadratic Forms. Recall in days past when you were given an equation which looked like

Differential Geometry. Peter Petersen

Principal Component Analysis (PCA) The Gaussian in D dimensions

Camera calibration. Outline. Pinhole camera. Camera projection models. Nonlinear least square methods A camera calibration tool

Computerized tomography. CT Reconstruction. Physics of beam attenuation. Physics of beam attenuation

SIMULATION OF TURBULENT FLOW AND HEAT TRANSFER OVER A BACKWARD-FACING STEP WITH RIBS TURBULATORS

FRTN10 Exercise 12. Synthesis by Convex Optimization

FEA Solution Procedure

Optical Flow, KLT Feature Tracker.

Transcription:

Featre etraction: Corners and blobs

Wh etract featres? Motiation: panorama stitching We hae two images how do we combine them?

Wh etract featres? Motiation: panorama stitching We hae two images how do we combine them? Step 1: etract featres Step : match featres

Wh etract featres? Motiation: panorama stitching We hae two images how do we combine them? Step 1: etract featres Step : match featres Step 3: align images

Characteristics of good featres Repeatabilit The same featre can be fond in seeral images despite geometric and photometric transformations Salienc ach featre has a distinctie description Compactness and efficienc Man fewer featres than image piels Localit A featre occpies a relatiel small area of the image; robst to cltter and occlsion

Applications Featre points are sed for: Motion tracking mage alignment 3D reconstrction Object recognition ndeing and database retrieal Robot naigation

Finding Corners Ke propert: in the region arond a corner image gradient has two or more dominant directions Corners are repeatable and distinctie CH i dmst h "A C bi d C d d D t t C.Harris and M.Stephens. "A Combined Corner and dge Detector. Proceedings of the 4th Ale Vision Conference: pages 147--151.

Corner Detection: Basic dea We shold easil recognize the point b looking throgh a small window Shifting a window in an direction shold gie a large change in intensit Sorce: A. fros flat region: no change in all directions edge : no change along the edge direction corner : significant change in all directions

Corner Detection: Mathematics Change in appearance for the shift []: [ ] = w Window fnction Shifted intensit ntensit Window fnction w = or 1 in window 0 otside Gassian Sorce: R. Szeliski

Corner Detection: Mathematics Change in appearance for the shift []: [ ] = w 00 3

Corner Detection: Mathematics Change in appearance for the shift []: [ ] = w We want to find ot how this fnction behaes for small shifts Second-order Talor epansion of abot 00 local qadratic approimation: 00 1 00 00 00 0 [ ] [ ] 00 00 00

Corner Detection: Mathematics [ ] w = Second-order Talor epansion of abot 00: 00 00 ] [ 1 00 ] [ 00 00 00 ] [ 00 ] [ 00 [ ] w = w = [ ] w w = [ ] w

Corner Detection: Mathematics [ ] w = Second-order Talor epansion of abot 00: 00 00 ] [ 1 00 ] [ 00 0 00 = 00 00 ] [ 00 ] [ 00 0 00 0 00 = = 00 00 w w = = 00 w =

Corner Detection: Mathematics [ ] w = Second-order Talor epansion of abot 00: w w ] [ w w ] [ 0 00 = 0 00 0 00 = = 00 00 w w = = 00 w =

Corner Detection: Mathematics The qadratic approimation simplifies to [ ] M where M is a second moment matri compted from image deriaties: es M = w M

nterpreting the second moment matri The srface is locall approimated b a qadratic form. Let s tr to nderstand its shape. M ] [ = w M = w M

nterpreting the second moment matri First consider the ais-aligned case gradients are either horizontal or ertical 0 λ gradients are either horizontal or ertical = = 1 0 0 λ λ w M f either λ is close to 0 then this is not a corner so look for locations where both are large.

nterpreting the second moment matri Consider a horizontal slice of : [ ] M = const This is the eqation of an ellipse.

nterpreting the second moment matri Consider a horizontal slice of : [ ] M = const This is the eqation of an ellipse. Diagonalization of M: M λ = R 1 0 1 R 0 λ The ais lengths of the ellipse are determined b the eigenales and the orientation is determined b R direction of the fastest change direction of the slowest change λ ma -1/ λ -1/ min

Visalization of second moment matrices

Visalization of second moment matrices

nterpreting the eigenales Classification of image points sing eigenales of M: λ dge λ >> λ 1 Corner λ 1 and λ are large λ 1 ~ λ ; increases in all directions λ 1 and λ are small; is almost constant in all directions Flat region dge λ 1 >> λ λ 1

Corner response fnction R = det M M α trace = λ1λ α λ1 λ α: constant t 0.04 04 to 0.06 06 dge R < 0 Corner R>0 R small Flat region dge R < 0

Harris detector: Steps 1. Compte Gassian deriaties at each piel. Compte second moment matri M in a Gassian window arond each piel 3. Compte corner response fnction R 4. Threshold R 5. Find local maima of response fnction nonmaimm sppression CH i d MSt h "A C bi d C d d D t t C.Harris and M.Stephens. "A Combined Corner and dge Detector. Proceedings of the 4th Ale Vision Conference: pages 147 151 1988.

Harris Detector: Steps

Harris Detector: Steps Compte corner response R

Harris Detector: Steps Find points with large corner response: R>threshold

Harris Detector: Steps Tk Take onl the points it of flocal lmaima of frr

Harris Detector: Steps

nariance and coariance We want featres to be inariant to photometric transformations and coariant to geometric transformations nariance: image is transformed and featres do not change Coariance: if we hae two transformed ersions of the same image featres shold be detected in corresponding locations

Models of mage Change Photometric Affine intensit change a b Geometric Rotation Scale Affine alid for: orthographic camera locall planar object

Affine intensit change Onl deriaties are sed => inariance to intensit shift b ntensit scale: a R threshold R image coordinate image coordinate Partiall inariant to affine intensit change

mage rotation llipse rotates bt its shape i.e. eigenales remains the same Corner response R is inariant w r t rotation and Corner response R is inariant w.r.t. rotation and corner location is coariant

Scaling Corner All points will be classified as edges Not inariant to scaling

Achieing scale coariance Goal: independentl detect corresponding regions in scaled ersions of the same image Need scale selection mechanism for finding characteristic region size that is coariant with the image transformation

Blob detection with scale selection

Recall: dge detection f dge d d g Deriatie of Gassian f d d g dge = maimm of deriatie Sorce: S. Seitz

dge detection Take f dge d d Second deriatie g of Gassian Laplacian d f d g dge = zero crossing of second deriatie Sorce: S. Seitz

From edges to blobs dge = ripple Blob = sperposition of two ripples maimm Spatial selection: the magnitde of the Laplacian response will achiee a maimm at the center of the blob proided the scale of the Laplacian is matched to the scale of the blob

Scale selection We want to find the characteristic scale of the blob b conoling it with Laplacians at seeral scales and looking for the maimm response Howeer Laplacian response decas as scale increases: original signal radis=8 increasing σ Wh does this happen?

Scale normalization The response of a deriatie of Gassian filter to a perfect step edge decreases as σ increases 1 σ π

Scale normalization The response of a deriatie of Gassian filter to a perfect step edge decreases as σ increases To keep response the same scale-inariant mst mltipl Gassian deriatie b σ Laplacian is the second Gassian deriatie so it mst be mltiplied b σ

ffect of scale normalization Original signal Unnormalized Laplacian response Scale-normalized Laplacian response maimm

Blob detection in D Laplacian of Gassian: Circlarl smmetric operator for blob detection in D g g g =

Blob detection in D Laplacian of Gassian: Circlarl smmetric operator for blob detection in D Scale-normalized: g g g = σ norm

Scale selection At what scale does the Laplacian achiee a maimm response to a binar circle of radis r? r image Laplacian

Scale selection At what scale does the Laplacian achiee a maimm response to a binar circle of radis r? To get maimm response the zeros of the Laplacian hae to be aligned with the circle The Laplacian is gien b p to scale: σ Therefore the maimm response occrs at e / σ σ = r /. r circle image Laplacian

Characteristic scale We define the characteristic scale of a blob as the scale that prodces peak of Laplacian response in the blob b center characteristic scale T. Lindeberg 1998. "Featre detection with atomatic scale selection." nternational Jornal of Compter Vision 30 : pp 77--116.

Scale-space blob detector 1. Conole image with scale-normalized Laplacian at seeral scales. Find maima of sqared Laplacian response in scale-space

Scale-space blob detector: ample

Scale-space blob detector: ample

Scale-space blob detector: ample

fficient implementation Approimating the Laplacian with a difference of Gassians: L= G G σ σ σ Laplacian DG DoG = G k σ G σ Difference of Gassians

fficient implementation Daid G. Lowe. "Distinctie image featres from scale-inariant kepoints. JCV 60 pp. 91-110 004.

nariance and coariance properties Laplacian blob response is inariant w.r.t. rotation and scaling Blob location is coariant w.r.t. rotation and scaling What abot intensit change?

Achieing affine coariance Consider the second moment matri of the window containing the blob: M = λ 1 1 w = R 0 0 R λ Recall: [ ] M = const direction of the fastest change λ ma -1/ λ -1/ min direction of the slowest change This ellipse isalizes the characteristic shape of the window

Affine adaptation eample Scale-inariant regions blobs

Affine adaptation eample Affine-adapted blobs

Affine adaptation Problem: the second moment window determined b weights w mst match the characteristic ti shape of the region Soltion: iteratie approach Use a circlar window to compte second moment matri Perform affine adaptation to find an ellipse-shaped window Recompte second moment matri sing new window and iterate

teratie affine adaptation K. Mikolajczk and C. Schmid Scale and Affine inariant interest point detectors JCV 601:63-86 004. http://www.robots.o.ac.k/~gg/research/affine/

Affine coariance Affinel transformed ersions of the same neighborhood will gie rise to ellipses that are related b the same transformation What to do if we want to compare these image regions? Affine normalization: transform these regions into same-size circles

Affine normalization Problem: There is no niqe transformation from an ellipse to a nit circle We can rotate or flip a nit circle and it still stas a nit circle

liminating rotation ambigit To assign a niqe orientation to circlar image windows: Create histogram of local l gradient directions in the patch Assign canonical orientation at peak of smoothed histogram 0 π

From coariant regions to inariant featres tract affine regions Normalize regions liminate rotational ambigit Compte appearance descriptors SFT Lowe 04

nariance s. coariance nariance: featrestransformimage = featresimage Coariance: featrestransformimage = transformfeatresimage Coariant detection => inariant description