Scribd
Upload
5 views

317-Epipolar Geometry-Intro

This document is a transcript of a lecture on stereo vision, specifically focusing on epipolar geometry. It discusses the concept of disparity, how it relates to depth perception, and the process of finding correspondences between two images based on camera alignment. The lesson sets the stage for understanding constraints in locating points in stereo images for depth calculation.

Uploaded by

saeb2saeb
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
5 views

317-Epipolar Geometry-Intro

This document is a transcript of a lecture on stereo vision, specifically focusing on epipolar geometry. It discusses the concept of disparity, how it relates to depth perception, and the process of finding correspondences between two images based on camera alignment. The lesson sets the stage for understanding constraints in locating points in stereo images for depth calculation.

Uploaded by

saeb2saeb
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

1

00:00:00,350 --> 00:00:02,210


Welcome back to Computer Vision.

2
00:00:02,210 --> 00:00:05,570
This is going to be the second in a series about stereo.

3
00:00:05,570 --> 00:00:09,745
In particular, we're going to focus on what's called epipolar geometry.

4
00:00:09,745 --> 00:00:12,500
We'd said last time that if we had two images and

5
00:00:12,500 --> 00:00:16,840
we knew something about the cameras, if we could find correspondences given that

6
00:00:16,840 --> 00:00:19,280
known relationship, we could find the depth.

7
00:00:19,280 --> 00:00:21,490
And we talked about disparity.

8
00:00:21,490 --> 00:00:26,080
And disparity was the idea that the location of a point in an image would

9
00:00:26,080 --> 00:00:31,020
change, depending upon the depth of the scene as I move my camera.

10
00:00:31,020 --> 00:00:34,840
And if I actually had the disparity I could make a disparity map,

11
00:00:34,840 --> 00:00:37,410
which tells me how all the points have shifted.

12
00:00:37,410 --> 00:00:41,560
And disparity was inversely proportional to the depth, and

13
00:00:41,560 --> 00:00:44,990
that's was this disparity map on the right is.

14
00:00:44,990 --> 00:00:47,710
But, how do we find the disparity?

15
00:00:47,710 --> 00:00:51,870
Essentially given one point in the image, we need to search for it in the other.
16
00:00:51,870 --> 00:00:56,710
But our intuition tells us, if I know something about how the two

17
00:00:56,710 --> 00:01:01,300
cameras are aligned, you give me one point in the picture on the left.

18
00:01:01,300 --> 00:01:04,519
It can't be anywhere on the picture on the right,

19
00:01:04,519 --> 00:01:09,400
it has to be somehow constrained in terms of where it can be on the left.

20
00:01:09,400 --> 00:01:12,900
This lesson is going to be about those constraints, and

21
00:01:12,900 --> 00:01:15,870
then we're going to assume those constraints when we go about actually finding

22
00:01:15,870 --> 00:01:15,870
the points in the next lesson.

You might also like