Scribd
Upload
14 views2 pages

301-Stereo Geometry-Intro

The lecture introduces stereo geometry, focusing on the relationship between camera geometry and scene geometry. It emphasizes the importance of stereo views in recovering depth information and sets the stage for further exploration of how multiple views relate to each other and the scene. The discussion includes examples of stereo pairs and depth maps, highlighting the computational aspects of these relationships.

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
14 views2 pages

301-Stereo Geometry-Intro

The lecture introduces stereo geometry, focusing on the relationship between camera geometry and scene geometry. It emphasizes the importance of stereo views in recovering depth information and sets the stage for further exploration of how multiple views relate to each other and the scene. The discussion includes examples of stereo pairs and depth maps, highlighting the computational aspects of these relationships.

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

00:00:00,480 --> 00:00:02,870


Hello, welcome back to computer vision.

2
00:00:02,870 --> 00:00:07,070
Today, we're going to start talking about, stereo and

3
00:00:07,070 --> 00:00:09,550
this particular lecture is on stereo geometry.

4
00:00:09,550 --> 00:00:12,260
In general, the next few lectures are going to be on

5
00:00:12,260 --> 00:00:16,760
the relationship between sort of camera geometry and scene geometry.

6
00:00:16,760 --> 00:00:20,510
We're starting with stereo because frankly you guys are going to do this,

7
00:00:20,510 --> 00:00:23,480
in a problem set and I want to be able to get you there quickly, and

8
00:00:23,480 --> 00:00:27,280
also in some sense it is the conceptually easiest way to get started.

9
00:00:27,280 --> 00:00:32,280
So stereo is really just a special case of having multiple views of an image,

10
00:00:32,280 --> 00:00:36,630
two particular views, and here's just an example of two views of a vase.

11
00:00:36,630 --> 00:00:39,900
In fact, we're going to get to this particular figure, oh,

12
00:00:39,900 --> 00:00:42,110
a little bit later when we talk about epipolar geometry.

13
00:00:42,110 --> 00:00:45,000
But the idea is that there's this relationship between two views, and

14
00:00:45,000 --> 00:00:47,940
you can use it to recover the depth that's there.

15
00:00:47,940 --> 00:00:51,720
We'll also take a look at sort of having arbitrary different views, and this is
16
00:00:51,720 --> 00:00:56,440
an image that comes from work on using special features to align objects and

17
00:00:56,440 --> 00:00:58,290
be able to detect that they're there, but, again,

18
00:00:58,290 --> 00:01:01,270
you have two different views and you have to figure out how they relate.

19
00:01:01,270 --> 00:01:03,960
This is an example of what you can get from stereos, so

20
00:01:03,960 --> 00:01:07,720
the picture on the left is one half of a stereo pair,

21
00:01:07,720 --> 00:01:12,540
on the right is a crude depth map, the brighter things are, the closer they are,

22
00:01:12,540 --> 00:01:16,310
in fact, we're going to talk about how to compute exactly that.

23
00:01:16,310 --> 00:01:18,770
And in general, you can have lots of views,

24
00:01:18,770 --> 00:01:22,250
where you have images taken from a whole bunch of places, or

25
00:01:22,250 --> 00:01:25,610
a whole bunch of directions, and it's a question of how do these images relate.

26
00:01:25,610 --> 00:01:27,360
And this notion of how images relate and

27
00:01:27,360 --> 00:01:27,360
how do they relate to the scene, that's what we're going to be focusing on.

You might also like