You are viewing the course site for a past offering of this course. The current offering may be found here.
Lecture 21: Image Sensors (90)

Here's a link to a more in depth article describing noise reduction by image averaging ( The discussion is pretty interesting, as it examines changes on a more specific level, and also demonstrates how rudimentary versions of it can be accomplished in easily accessible apps like photoshop!


While this does decrease noise for stationary subjects, the shutter speed for each frame is fixed for an exposure. Wouldn't this lead to an increase in aliasing effects due to motion? One way around this might be to have multiple sensors or to somehow offset the exposure time on the same sensor (kind of like pipelining reads), but I'm not sure how that'd work in terms of the electronics.


I think the downside of this technique with regards to capturing objects in motion was discussed in lecture.

On another note, I think I've seen a variation of this technique used to create photos of tourist attractions without any people. By using multiple exposures over a longer period of time (say, 1 photo every 30 seconds for 15 minutes) and then taking the median value of each pixel, you can supposedly get good results. I haven't had the patience to try it myself though. Here's an article with an example:


Is there any reason why someone would choose this instead of something like a median filter? Seems like the median filter would be pretty successful at removing noise without having to do with the side effects of blurring.


Done correctly, noise can actually add quite a bit of artistic feel to a photograph. However, done incorrectly it is quite distracting and generally degrades from the photo. It's interesting how subjective it is and how hard it is to find the proper balance.

You must be enrolled in the course to comment