You are viewing the course site for a past offering of this course. The current offering may be found here.
Lecture 22: Image Processing (42)
andrewdcampbell

Median filtering is particularly strong-suited for removing speckle noise / salt-and-pepper noise. I learned this when I was trying to clean up an image rendered from our pathtracing project to use as wallpaper; out of all the filtering techniques I tried (e.g. Gaussian, box, etc.), none could remove the speckles while still preserving the edge quality.

fywu85

Out of curiosity, I wonder why choosing median of the filter? What is the underlying rationale for this? For instances, what will happen if we use 75 percentile? What will happen if we take the max?

yzyz

You want to use median because it is the most representative of the "true" value of the pixel, assuming that the noise isn't biased in a particular direction.

You must be enrolled in the course to comment