lec9 prelim

Lec9 prelim - Lecture 9 HW2 due today HW3 to be posted later today Due Wednesday Oct 6 Office Hours today only 1:30 to 2:30 PM Last Class Line and

Info iconThis preview shows pages 1–7. Sign up to view the full content.

View Full Document Right Arrow Icon
USC CS574: Computer Vision, Fall 2010 Copyright 2010, by R. Nevatia 1 Lecture 9: Sept 22, 10 • HW2 due today • HW3 to be posted later today, Due Wednesday Oct 6 • Office Hours, today only: 1:30 to 2:30 PM • Last Class – Line and curve fitting – Corner features –S IFT fea tu re s • Updated tutorial posted with example including an affine transformation • Today’s objective – Intro to Stereo
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
USC CS574: Computer Vision, Fall 2010 Copyright 2010, by R. Nevatia 2 GLOH: Modified Version of SIFT • Gradient Location Orientation Histograms: uses Log-polar bins, see RS book, fig. 4.19. Claims to have somewhat better performance than SIFT
Background image of page 2
USC CS574: Computer Vision, Fall 2010 Copyright 2010, by R. Nevatia 3 Spatio-Temporal Features • Features can be generalized to videos • Consider video to span a volume in (x,y,t) space • Find distinctive features in this volume • Descriptors can include not only distribution of spatial gradients but also temporal gradients. • We will look into more details towards the end of the course if time permits.
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
USC CS574: Computer Vision, Fall 2010 Copyright 2010, by R. Nevatia 4 Inference of 3-D • From a single image – Intensity (shape from shading) – Texture (requires regular texture, not covered in class/book) – 2-D shape (details not covered in class/book) • From multiple images, static scene – From multiple fixed cameras or from camera motion – Stereo: Two fixed cameras Structure from motion : Estimate camera motion and 3-D scene structure • From object motion, stationary camera – Must infer object motion in addition to 3-D surface • Most difficult case is when both the sensor and some of the objects in the scene move
Background image of page 4
USC CS574: Computer Vision, Fall 2010 Copyright 2010, by R. Nevatia 5 Steps of Stereo Analysis • Point correspondence between views – Difficulties/details covered later • Triangulation – Assuming calibrated cameras, each view provides a ray, say R and R', on which the 3-D point must lie: compute intersection
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
USC CS574: Computer Vision, Fall 2010 Copyright 2010, by R. Nevatia
Background image of page 6
Image of page 7
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 11/23/2010 for the course CS 574 taught by Professor Ramnevatia during the Fall '10 term at USC.

Page1 / 22

Lec9 prelim - Lecture 9 HW2 due today HW3 to be posted later today Due Wednesday Oct 6 Office Hours today only 1:30 to 2:30 PM Last Class Line and

This preview shows document pages 1 - 7. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online