May 2004Stereo1 Introduction to Computer Vision CS / ECE 181B Tuesday, May 11, 2004  Multiple view geometry and stereo  Handout #6 available (check with.

Slides:



Advertisements
Similar presentations
Epipolar Geometry.
Advertisements

Lecture 11: Two-view geometry
3D reconstruction.
MASKS © 2004 Invitation to 3D vision Lecture 7 Step-by-Step Model Buidling.
Two-View Geometry CS Sastry and Yang
Jan-Michael Frahm, Enrique Dunn Spring 2012
Two-view geometry.
Lecture 8: Stereo.
Dr. Hassan Foroosh Dept. of Computer Science UCF
Camera calibration and epipolar geometry
Structure from motion.
Computer Vision : CISC 4/689
Epipolar Geometry Class 7 Read notes Feature tracking run iterative L-K warp & upsample Tracking Good features Multi-scale Transl. Affine.
Epipolar geometry. (i)Correspondence geometry: Given an image point x in the first view, how does this constrain the position of the corresponding point.
Structure from motion. Multiple-view geometry questions Scene geometry (structure): Given 2D point matches in two or more images, where are the corresponding.
Uncalibrated Geometry & Stratification Sastry and Yang
Lecture 21: Multiple-view geometry and structure from motion
Multiple View Geometry
Multiple-view Reconstruction from Points and Lines
Multiple View Geometry Marc Pollefeys University of North Carolina at Chapel Hill Modified by Philippos Mordohai.
Lecture 20: Two-view geometry CS6670: Computer Vision Noah Snavely.
3D Computer Vision and Video Computing 3D Vision Lecture 14 Stereo Vision (I) CSC 59866CD Fall 2004 Zhigang Zhu, NAC 8/203A
Projected image of a cube. Classical Calibration.
Lec 21: Fundamental Matrix
CSE473/573 – Stereo Correspondence
Multiple View Geometry
Multiple View Geometry. THE GEOMETRY OF MULTIPLE VIEWS Reading: Chapter 10. Epipolar Geometry The Essential Matrix The Fundamental Matrix The Trifocal.
3-D Scene u u’u’ Study the mathematical relations between corresponding image points. “Corresponding” means originated from the same 3D point. Objective.
Epipolar geometry Class 5. Geometric Computer Vision course schedule (tentative) LectureExercise Sept 16Introduction- Sept 23Geometry & Camera modelCamera.
Multi-view geometry. Multi-view geometry problems Structure: Given projections of the same 3D point in two or more images, compute the 3D coordinates.
Automatic Camera Calibration
Computer vision: models, learning and inference
Lecture 11 Stereo Reconstruction I Lecture 11 Stereo Reconstruction I Mata kuliah: T Computer Vision Tahun: 2010.
Multi-view geometry.
Lecture 12 Stereo Reconstruction II Lecture 12 Stereo Reconstruction II Mata kuliah: T Computer Vision Tahun: 2010.
1 Preview At least two views are required to access the depth of a scene point and in turn to reconstruct scene structure Multiple views can be obtained.
Projective cameras Motivation Elements of Projective Geometry Projective structure from motion Planches : –
Course 12 Calibration. 1.Introduction In theoretic discussions, we have assumed: Camera is located at the origin of coordinate system of scene.
Lecture 04 22/11/2011 Shai Avidan הבהרה : החומר המחייב הוא החומר הנלמד בכיתה ולא זה המופיע / לא מופיע במצגת.
3D Sensing and Reconstruction Readings: Ch 12: , Ch 13: , Perspective Geometry Camera Model Stereo Triangulation 3D Reconstruction by.
Multiview Geometry and Stereopsis. Inputs: two images of a scene (taken from 2 viewpoints). Output: Depth map. Inputs: multiple images of a scene. Output:
Lec 22: Stereo CS4670 / 5670: Computer Vision Kavita Bala.
Announcements Project 3 due Thursday by 11:59pm Demos on Friday; signup on CMS Prelim to be distributed in class Friday, due Wednesday by the beginning.
EECS 274 Computer Vision Stereopsis.
Geometry of Multiple Views
Computer Vision Lecture #10 Hossam Abdelmunim 1 & Aly A. Farag 2 1 Computer & Systems Engineering Department, Ain Shams University, Cairo, Egypt 2 Electerical.
Single-view geometry Odilon Redon, Cyclops, 1914.
CSE 185 Introduction to Computer Vision Stereo. Taken at the same time or sequential in time stereo vision structure from motion optical flow Multiple.
Two-view geometry. Epipolar Plane – plane containing baseline (1D family) Epipoles = intersections of baseline with image planes = projections of the.
stereo Outline : Remind class of 3d geometry Introduction
Feature Matching. Feature Space Outlier Rejection.
Computer vision: models, learning and inference M Ahad Multiple Cameras
Correspondence and Stereopsis Original notes by W. Correa. Figures from [Forsyth & Ponce] and [Trucco & Verri]
Geometry Reconstruction March 22, Fundamental Matrix An important problem: Determine the epipolar geometry. That is, the correspondence between.
Stereo March 8, 2007 Suggested Reading: Horn Chapter 13.
Structure from motion Multi-view geometry Affine structure from motion Projective structure from motion Planches : –
Lec 26: Fundamental Matrix CS4670 / 5670: Computer Vision Kavita Bala.
Correspondence and Stereopsis. Introduction Disparity – Informally: difference between two pictures – Allows us to gain a strong sense of depth Stereopsis.
CSE 185 Introduction to Computer Vision Stereo 2.
Multi-view geometry. Multi-view geometry problems Structure: Given projections of the same 3D point in two or more images, compute the 3D coordinates.
55:148 Digital Image Processing Chapter 11 3D Vision, Geometry
CS4670 / 5670: Computer Vision Kavita Bala Lec 27: Stereo.
Two-view geometry Computer Vision Spring 2018, Lecture 10
Epipolar geometry.
3D Photography: Epipolar geometry
Reconstruction.
Two-view geometry.
Two-view geometry.
Multi-view geometry.
Chapter 11: Stereopsis Stereopsis: Fusing the pictures taken by two cameras and exploiting the difference (or disparity) between them to obtain the depth.
Presentation transcript:

May 2004Stereo1 Introduction to Computer Vision CS / ECE 181B Tuesday, May 11, 2004  Multiple view geometry and stereo  Handout #6 available (check with Isabelle) Ack: M. Turk and M. Pollefeys

May 2004Stereo2 Midterm

May 2004Stereo3 Seeing in 3D Humans can perceive depth, shape, etc. – 3D properties of the world –How do we do it? We use many cues –Oculomotor convergence/divergence –Accomodation (changing focus) –Motion parallax (changing viewpoint) –Monocular depth cues  Occlusion, perspective, texture gradients, shading, size –Binocular disparity (stereo) How can computers perceive depth?

May 2004Stereo4

May 2004Stereo5

May 2004Stereo6 Multiple views and depth

May 2004Stereo7 Why multiple views? A camera projects the 3D world into 2D images This is not always a problem – humans can figure out a lot from a 2D view!

May 2004Stereo8 Why multiple views? But precise 3D information (distance, depth, shape, curvature, etc.) is difficult or impossible to obtain from a single view In order to measure distances, sizes, angles, etc. we need multiple views (and calibrated cameras!) –Monocular  binocular  trinocular… C1C1 C2C2 C3C3

May 2004Stereo9 Multiple view geometry C1C1 C2C2 C3C3 Two big questions for multiple view geometry problems: –Which are possible? –Which are most likely? There are many possible configurations of scene points that could have created corresponding points in multiple views

May 2004Stereo10 Questions Correspondence geometry: Given an image point x in the first view, how does this constrain the position of the corresponding point x’ in the second image? Camera geometry (motion): Given a set of corresponding image points {x i ↔x’ i }, i=1,…,n, what are the cameras P and P’ for the two views? Scene geometry (structure): Given corresponding image points x i ↔x’ i and cameras P, P’, what is the position of (their pre-image) X in space? M. Pollefeys

May 2004Stereo11 Two-view geometry C1C1 C2C2 Epipolar line Not necessarily along a row of the image p The epipolar geometry is defined by the origins of the camera coordinate frames, the scene point P, and the locations of the image planes

May 2004Stereo12 C,C’,x,x’ and X are coplanar Epipolar geometry

May 2004Stereo13 What if only C,C’,x are known? Epipolar Geometry

May 2004Stereo14 All points on  project on l and l’ Epipolar Geometry

May 2004Stereo15 Family of planes  and lines l and l’ Intersection in e and e’ Epipolar Geometry

May 2004Stereo16 epipoles e,e’ = intersection of baseline with image plane = projection of projection center in other image = vanishing point of camera motion direction an epipolar plane = plane containing baseline (1-D family) an epipolar line = intersection of epipolar plane with image (always come in corresponding pairs) Epipolar geometry

May 2004Stereo17 Epipolar geometry Epipolar Plane Epipoles Epipolar Lines Baseline C1C1 C2C2

May 2004Stereo18 Epipolar constraint Potential matches for p have to lie on the corresponding epipolar line l’ Potential matches for p’ have to lie on the corresponding epipolar line l

May 2004Stereo19 Example: converging cameras

May 2004Stereo20 Example: motion parallel with image plane

May 2004Stereo21 Example: forward motion e e’

May 2004Stereo22 Trinocular epipolar constraint

May 2004Stereo23 Basic approach to stereo vision Find features of interest in N image views –The “correspondence problem” Triangulate –A method to measure distance and direction by forming a triangle and using trigonometry Reconstruct object/scene depth –From dense points –From sparse points

May 2004Stereo24 Step 1: The correspondence problem Given a “point” in one image, find the location of that same point in a second image (and maybe third, and fourth, …) p A search problem: Given point p in the left image, where in the right image should we search for a corresponding point? p’ Sounds easy, huh?

May 2004Stereo25 Correspondence problem Right imageLeft image What is a point? How do we compare points in different images? (Similarity measure)

May 2004Stereo26 Correspondence problem Left imageRight image

May 2004Stereo27 The correspondence problem A classically difficult problem in computer vision –Is every point visible in both images? –Do we match points or regions or …? –Are corresponding (L-R) image regions similar? Correspondence is easiest when the depth is large compared with the camera baseline distance –Because the cameras then have about the same viewpoint –But… Two classes of stereo correspondence algorithms: –Feature based (sparse) – corners, edges, lines, … –Correlation based (dense)  How large a window of support to use?

May 2004Stereo28 Multiple views What do you need to know in order to calculate the depth (or location) of the point that causes p and p' ? C1C1 C2C2 p p Values of p = (u, v) and p = (u, v) Locations of C 1 and C 2 (full extrinsic parameters) –Rigid transformation between C 1 and C 2 Intrinsic parameters of C 1 and C 2

May 2004Stereo29 Duality: Calibration and stereo Given calibrated cameras, we can find depth of points Given corresponding points, we can calibrate the cameras C1C1 C2C2 C1C1 C2C2

May 2004Stereo30 Example: Extrinsic parameters from 3 points C1C1 C2C2 1 known point 2 known points 3 known points In this case, we know the point correspondences and the point distances. If we only know the correspondences, we’ll need at least five points

May 2004Stereo31 The geometry of multiple views Epipolar Geometry –The Essential Matrix –The Fundamental Matrix The Trifocal Tensor The Quadrifocal Tensor Baseline cc’

May 2004Stereo32 Epipolar geometry Epipolar Plane Epipoles Epipolar Lines Baseline C1C1 C2C2

May 2004Stereo33 Epipolar constraint Potential matches for p have to lie on the corresponding epipolar line l’ Potential matches for p’ have to lie on the corresponding epipolar line l

May 2004Stereo34 Epipolar lines example

May 2004Stereo35 Matrix form of cross product The cross product of two vectors is a third vector, perpendicular to the others (right hand rule)

May 2004Stereo36 p p Case 1: Calibrated camera O O P OP Op O P O p OO Op · (OO  O p ) = ? Op · (OO  O p ) = 0 [ R t ] – rigid trans. from O to O p · (t  Rp ) = 0 This can be written in matrix form as: p T E p = 0

May 2004Stereo37 Essential Matrix p p O O P OP Op O P O p OO p T E p = 0 p · (t  Rp ) = 0 E - Essential Matrix

May 2004Stereo38 The Essential Matrix E describes the transformation between camera coordinate frames E has five degrees of freedom –Defined up to a scale factor, since p T E p = 0 Why only five? –A rigid transformation has six degrees of freedom 3 rotation parameters, 2 translation direction parameters –Why only translation direction?

May 2004Stereo39

May 2004Stereo40 “Up to a scale factor” This is always the case with camera calibration and stereo –Shrink everything 10x and it all looks the same! Typically there is something we know that we can use to specify the scale factor –E.g., the baseline, the size of an object, the depth of a point/plane

May 2004Stereo41 Camera calibration from E With five unknowns, theoretically we can recover the essential matrix E by writing p T E p = 0 for five corresponding pairs of points –5 equations and 5 unknowns –We don’t need to know anything about the points (e.g., their depth), only that they project to p i and p i –There are, however, limitations… This is used for camera calibration (extrinsic parameters) C1C1 C2C2

May 2004Stereo42 Case 2: Uncalibrated camera Intrinsic parameters not known Fundamental Matrix Points in the normalized image plane

May 2004Stereo43 geometric derivation mapping from 2-D to 1-D family (rank 2) Fundamental Matrix F

May 2004Stereo44 The Fundamental Matrix F has seven independent parameters A simple, linear technique to recover F from corresponding point locations is the “eight point algorithm” From F, we can recover the epipolar geometry of the cameras –Not saying how… This is called weak calibration

May 2004Stereo45 The eight-point algorithm Invert and solve for F

May 2004Stereo46 Least squares approach Minimize: under the constraint |F| 2 = 1 If n > 8

May 2004Stereo47 Nonlinear least-squares approach Minimize with respect to the coefficients of F Point in image 1 Epipolar line in image 1 caused by p Nonlinear – initialize it from the results of the eight-point algorithm

May 2004Stereo48 Least squares 8-point algorithmHartley’s normalized 8-point alg.

May 2004Stereo49 Stereo vision (Stereopsis)

May 2004Stereo50 I1I2I10

May 2004Stereo51 Basic approach to stereo vision Find features of interest in N image views –The “correspondence problem” Triangulate –A method to measure distance and direction by forming a triangle and using trigonometry Reconstruct object/scene depth –From dense points –From sparse points

May 2004Stereo52 1. Correspondence C1C1 C2C2 2. Triangulation 3. Reconstruction

May 2004Stereo53 Problem… Measurement error causes point Q to be seen at location p rather than the correct location q –A least squares method will triangulate to point P

May 2004Stereo54 Correspondence Knowing the epipolar geometry certainly helps –Look on (and near) the epipolar line But correspondence is hard! Two approaches –Try to improve correspondence matching –Try to avoid correspondence matching C1C1 C2C2 p

May 2004Stereo55 Image rectification Stereo calculations can be much simplified if the two images are rectified – replaced by two equivalent images with a common image plane parallel to the baseline Single, common image plane Epipolar lines are image scan lines

May 2004Stereo56 Rectification example

May 2004Stereo57 Correlation based stereo matching Texture-mapped reconstructed surface

May 2004Stereo58 Multiscale edge-based stereo Discussion session Friday/Monday One of the two input images Laplacian filtering at four scales Zero crossings

May 2004Stereo59 Depth map Reconstructed surface