MASTERCLASS

# How images are formed

### Lessons

#### Transcript

So far we've just considered the case of perspective projection. It's sort of image that is formed by a simple lens and the same sort of optics that we have in our own eyes. But there are many different sorts of cameras and different sorts of lenses.

They don't necessarily perform perspective projection. In this particular case, we're looking at a fish eye lens image. A fish eye lens here has got a very very wide field of view. If we look at the image captured by the fish eye lens, we notice a few things. Most importantly we note that the straightness of line is not preserved. So we see on top of the laptop on the side of the coke can, these are no longer straight lines. They appear to be curves because this is not a perspective projection. Lenses work by refraction. That is a bend of rays of light that are passing through them.

Another way we can do this is by reflection. We use curved mirrors instead of lenses in order to bend the light rays. Here is the example again from other journeys of Isaac Newton of his early reflecting telescope. Sometimes very very big lenses are actually made by combinations of shaped mirrors and lenses.

These sort of hybrid imaging systems are referred to as catadioptrics. This is derived from the Greek word dioptrics for refraction and catoptrics for reflection.

Another type of imaging system again on this catadioptric model is a panoramic lens. You can see it quite clearly here. What we have down the bottom is a fairly standard camera with a lens in front of it. And then there is a stalk which holds a parabolic mirror. So the camera is actually looking at the parabolic mirror.

This arrangement allows us to create panoramas. So the raw panorama, what the camera actually sees, is shown here on the bottom right. We see a very very distorted view of the world. Because we know the mathematics of this image formation process we can unwrap the image and that is what we use to create the panorama shown at the top.

What we note with this panoramic image once again is that straight lines are not preserved. They appear to be very very seriously distorted.

The pinhole or lensed camera is very similar to our eye, but there are lots of other ways to build a camera.

### Professor Peter Corke

Professor of Robotic Vision at QUT and Director of the Australian Centre for Robotic Vision (ACRV). Peter is also a Fellow of the IEEE, a senior Fellow of the Higher Education Academy, and on the editorial board of several robotics research journals.

### Skill level

This content assumes an understanding of high school-level mathematics, e.g. trigonometry, algebra, calculus, physics (optics) and some knowledge/experience of programming (any language).