Announcements. The equation of projection. Image Formation and Cameras

Similar documents
Announcements. Equation of Perspective Projection. Image Formation and Cameras

Cameras and Radiometry. Last lecture in a nutshell. Conversion Euclidean -> Homogenous -> Euclidean. Affine Camera Model. Simplified Camera Models

Announcements. Image Formation: Outline. Homogenous coordinates. Image Formation and Cameras (cont.)

COSC579: Scene Geometry. Jeremy Bolton, PhD Assistant Teaching Professor

Rigid Body Motion and Image Formation. Jana Kosecka, CS 482

Homogeneous Coordinates. Lecture18: Camera Models. Representation of Line and Point in 2D. Cross Product. Overall scaling is NOT important.

Vision Review: Image Formation. Course web page:

MAN-522: COMPUTER VISION SET-2 Projections and Camera Calibration

Camera Model and Calibration. Lecture-12

Image formation. Thanks to Peter Corke and Chuck Dyer for the use of some slides

Camera model and multiple view geometry

Discriminant Functions for the Normal Density

Geometric Model of Camera

Camera Model and Calibration

Robotics - Projective Geometry and Camera model. Marcello Restelli

CS6670: Computer Vision

3D Geometry and Camera Calibration

Image formation - About the course. Grading & Project. Tentative Schedule. Course Content. Students introduction

Image Formation. Antonino Furnari. Image Processing Lab Dipartimento di Matematica e Informatica Università degli Studi di Catania

Lecture 11 MRF s (conbnued), cameras and lenses.

CSE 252B: Computer Vision II

Representing the World

How to achieve this goal? (1) Cameras

3D Sensing. Translation and Scaling in 3D. Rotation about Arbitrary Axis. Rotation in 3D is about an axis

521466S Machine Vision Exercise #1 Camera models

CS4670: Computer Vision

Modeling Transformations

Perspective projection and Transformations

Geometric camera models and calibration

Robot Vision: Camera calibration

Camera Models and Image Formation. Srikumar Ramalingam School of Computing University of Utah

Agenda. Rotations. Camera models. Camera calibration. Homographies

Part Images Formed by Flat Mirrors. This Chapter. Phys. 281B Geometric Optics. Chapter 2 : Image Formation. Chapter 2: Image Formation

Chapter 36. Image Formation

Lecture 8: Camera Models

3D Geometry and Camera Calibration

Modeling Transformations

An introduction to 3D image reconstruction and understanding concepts and ideas

ECE-161C Cameras. Nuno Vasconcelos ECE Department, UCSD

3D Transformations. CS 4620 Lecture Kavita Bala w/ prior instructor Steve Marschner. Cornell CS4620 Fall 2015 Lecture 11

3-D D Euclidean Space - Vectors

Remember: The equation of projection. Imaging Geometry 1. Basic Geometric Coordinate Transforms. C306 Martin Jagersand

Computer Vision Project-1

Perspective Projection Transformation

Computer Vision cmput 428/615

3D Vision Real Objects, Real Cameras. Chapter 11 (parts of), 12 (parts of) Computerized Image Analysis MN2 Anders Brun,

Announcements. Rotation. Camera Calibration

Laser sensors. Transmitter. Receiver. Basilio Bona ROBOTICA 03CFIOR

Computer Vision and Applications. Prof. Trevor. Darrell. Class overview Administrivia & Policies Lecture 1

Announcements. Camera Calibration. Thin Lens: Image of Point. Limits for pinhole cameras. f O Z

Computer Vision. Coordinates. Prof. Flávio Cardeal DECOM / CEFET- MG.

3D Transformations. CS 4620 Lecture 10. Cornell CS4620 Fall 2014 Lecture Steve Marschner (with previous instructors James/Bala)

Projective Geometry and Camera Models

Lecture PowerPoint. Chapter 25 Physics: Principles with Applications, 6 th edition Giancoli

Computer Vision Projective Geometry and Calibration

Autonomous Navigation for Flying Robots

3-Dimensional Viewing

(x, y) (ρ, θ) ρ θ. Polar Coordinates. Cartesian Coordinates

Introduction to Computer Vision. Introduction CMPSCI 591A/691A CMPSCI 570/670. Image Formation

Outline The Refraction of Light Forming Images with a Plane Mirror 26-3 Spherical Mirror 26-4 Ray Tracing and the Mirror Equation

DD2423 Image Analysis and Computer Vision IMAGE FORMATION. Computational Vision and Active Perception School of Computer Science and Communication

Image Formation. 2. Camera Geometry. Focal Length, Field Of View. Pinhole Camera Model. Computer Vision. Zoltan Kato

Camera Models and Image Formation. Srikumar Ramalingam School of Computing University of Utah

The image is virtual and erect. When a mirror is rotated through a certain angle, the reflected ray is rotated through twice this angle.

Homogeneous Coordinates

Outline. ETN-FPI Training School on Plenoptic Sensing

CS559: Computer Graphics

AP Physics: Curved Mirrors and Lenses

Agenda. Perspective projection. Rotations. Camera models

Ray Optics I. Last time, finished EM theory Looked at complex boundary problems TIR: Snell s law complex Metal mirrors: index complex

Pin Hole Cameras & Warp Functions

Refraction at a single curved spherical surface

Final Exam. Today s Review of Optics Polarization Reflection and transmission Linear and circular polarization Stokes parameters/jones calculus

Understanding Variability

Projective Geometry and Camera Models

Camera Projection Models We will introduce different camera projection models that relate the location of an image point to the coordinates of the

Waves & Oscillations

Refraction of Light. This bending of the ray is called refraction

Chapter 23. Geometrical Optics (lecture 1: mirrors) Dr. Armen Kocharian

Geometry of a single camera. Odilon Redon, Cyclops, 1914

CS4670: Computer Vision

Agenda. Rotations. Camera calibration. Homography. Ransac

Pinhole Camera Model 10/05/17. Computational Photography Derek Hoiem, University of Illinois

Visual Recognition: Image Formation

LIGHT & OPTICS. Fundamentals of Physics 2112 Chapter 34 1

Determining the 2d transformation that brings one image into alignment (registers it) with another. And

Viewing Transformations I Comp 535

CPSC 425: Computer Vision

Waves & Oscillations

Computer Vision Projective Geometry and Calibration. Pinhole cameras

Computer Vision Projective Geometry and Calibration. Pinhole cameras

Jorge Salvador Marques, geometric camera model

Geometry of image formation

3D Computer Vision II. Reminder Projective Geometry, Transformations. Nassir Navab. October 27, 2009

Exercise 12 Geometrical and Technical Optics WS 2013/2014

Assignment 2 : Projection and Homography

Transformations. Examples of transformations: shear. scaling

Computer Vision CS 776 Fall 2018

CMPSCI 670: Computer Vision! Image formation. University of Massachusetts, Amherst September 8, 2014 Instructor: Subhransu Maji

Cameras and Stereo CSE 455. Linda Shapiro

Transcription:

Announcements Image ormation and Cameras Introduction to Computer Vision CSE 52 Lecture 4 Read Trucco & Verri: pp. 5-4 HW will be on web site tomorrow or Saturda. Irfanview: http://www.irfanview.com/ is a good windows utilit for manipulating images. Tr v for linu. inhole Camera: erspective projection Abstract camera model - bo with a small hole in it Geometric Aspects of erspective rojection oints project to points Lines project to lines Angles & distances (or ratios) are NT preserved under perspective Vanishing point Image plane orsth&once The equation of projection Euclidean -> Homogenous-> Euclidean In 2-D Euclidean -> Homogenous: (, ) -> λ (,,) (can just take λ =) Homogenous -> Euclidean: (,, ) -> (/, /) Cartesian coordinates: We have, b similar triangles, that (,, ) -> (f /, f /, -f) Ignore the third coordinate, and get In 3-D Euclidean -> Homogenous: (,, ) -> λ(,,,) (can just take λ =) Homogenous -> Euclidean: (,,, w) -> (/w, /w, /w)

Turn The camera matri into homogenous coordinates HC s for 3D point are (X,Y,Z,) HC s for point in image are (U,V,W) Affine Camera Model Take erspective projection equation, and perform Talor Series Epansion about (some point (,, ). Drop terms of higher order than linear. Resulting epression is called affine camera model. roperties ts. map to pts, lines map to lines arallel lines map to parallel lines (no vanishing point at infinit) Ratios of distance/angles preserved rthographic projection Start with affine camera model, and take Talor series about (,, o )= (,, ) a point on optical ais What if camera coordinate sstem differs from object coordinate sstem {c} {W} Depth () is lost Euclidean Coordinate Sstems Coordinate Changes: ure Translations No rotation (e.g., i A =i B etc) B = B A + A, B = A + B A 2

Rotation Matri Coordinate Changes: ure Rotations i A.i B j A.i B k A i B B A R = i A j B j A.j B k A.j B = i A k B j A.k B k A k B A i B T A j B T A k B T = [ B i B A j B A k A ] A rotation matri R has the following properties: Coordinate Changes: Rigid Transformations Its inverse is equal to its transpose R - = R T Its determinant is equal to : det(r)=. r equivalentl: Rows (or columns) of R form a right-handed orthonormal coordinate sstem. Rotation: Homogenous Coordinates About ais ' ' ' = sin θ -sin θ rot(,θ) Note: coordinate doesn t change after rotation θ p' p About ais: About ais: ' ' ' ' ' ' = = Rotation -sin θ sin θ -sin θ sin θ 3

Roll-itch-Yaw Rotation Rotate(k, θ) Rotation b angle θ about (k, k, k), a unit vector (Rodrigues ormula) θ k Euler Angles ' ' ' = kk(-c)+c kk(-c)+ks kk(-c)-ks kk(-c)-ks kk(-c)+c kk(-c)-ks kk(-c)+ks kk(-c)-ks kk(-c)+c where c = & s = sin θ Homogeneous Representation of Rigid Transformations Transformation represented b 4 b 4 Matri Block Matri Multiplication Given A = A A 2 B = B B 2 A 2 A 22 B 2 B 22 What is AB? Camera parameters Issue camera ma not be at the origin, looking down the -ais etrinsic parameters (Rigid Transformation) one unit in camera coordinates ma not be the same as one unit in world coordinates intrinsic parameters - focal length, principal point, aspect ratio, angle between aes, etc. X U Transformation Transformation V = representing representing Y Z W intrinsic parameters etrinsic parameters T 3 3 4 4 Camera Calibration What about light?, estimate intrinsic and etrinsic camera parameters See tetbook for how to do it. 4

Getting more light Bigger Aperture Limits for pinhole cameras inhole Camera Images with Variable Aperture The reason for lenses 2 mm mm.6 mm.35 mm.5 mm.7 mm Thin Lens Thin Lens: Center ptical ais Rotationall smmetric about optical ais. Spherical interfaces. All ras that enter lens along line pointing at emerge in same direction. 5

Thin Lens: ocus Thin Lens: Image of oint Incoming light ras parallel to the optical ais pass through the focus, All ras passing through lens and starting at converge upon Thin Lens: Image of oint Thin Lens: Image lane Z f Z Q Q Image lane A price: Whereas the image of is in focus, the image of Q isn t. Thin Lens: Aperture ield of View Image lane Smaller Aperture -> Less Blur inhole -> No Blur Image lane f ield of View 6

Deviations from the lens model Deviations from this ideal are aberrations Two tpes. geometrical 2. chromatic spherical aberration astigmatism distortion coma Aberrations are reduced b combining lenses Spherical aberration Ras parallel to the ais do not converge uter portions of the lens ield smaller focal lenghts Compound lenses Distortion magnification/focal length different for different angles of inclination Chromatic aberration Inde of refraction of lens depends on wavelength of light pincushion (tele-photo) barrel (wide-angle) Can be corrected! (if parameters are know) Chromatic aberration Spatial Non-Uniformit Ras of different wavelengths focused in different planes Cannot be removed completel Sometimes achromatiation is achieved for more than 2 wavelengths camera Iris Litvinov & Schechner, radiometric nonidealities 7

Vignetting nl part of the light reaches the sensor eripher of the image is dimmer 8