Previous | Next --- Slide 45 of 70
Back to Lecture Thumbnails

Here's a link to a nice explanation of temporal aliasing. Here a video camera is sampling a time varying 2D signal (f(t) is an image) at the video camera's frame rate. When the image changes rapidly in time compared to the frame rate aliasing occurs.

Here's another video illustrating examples of the "wagon wheel" effect:

Question: Have you ever noticed temporal aliasing when looking at wheels with your own eyes during the day? What about at night?


To make this even more interesting, let's take a look at the "rolling shutter" effect (also noticeable in the example in the slide):

Rolling shutter effect

We think of "samples" as the value of the signal f(t) at a particular time t. A sample of a video signal would be a frame at a discrete time. In reality, however, everything takes time... and sampling is no exception. And this is the key to to rolling shutter effect. In the real world, it always takes some amount of time for the camera to take a sample of the video signal. Most of the cameras today that use CMOS sensors that have rolling shutters. Instead of taking the whole image at an instant, they rapidly scan the video signal in horizontal or vertical strips. When an object moves extremely fast across the frame, we will actually get aliasing in our samples!

Here's a video explaining the rolling shutter effect with some awesome examples.

Question: How does the rolling shutter effect differ from the wagon wheel effect? How are they related?


I guess the rolling shutter effect is more of an interference pattern between the rate of sampling and the natural frequency of the signal, whereas the rolling shutter effect is a a breakdown of the assumption that sampling takes infinitely small time.