What is Subsampling in image processing?

Question

I was going through some image processing document. And I came cross this term "sub sampling". What does it mean? Is it sampling again already sampled data .. ?

Answer 1

Subsampling does not necessarily mean resampling previously sampled data, although it often gets implemented that way in practice.

It really refers to sampling at a rate (either in space or time) that's lower than the Nyquist criterion would indicate. It usually follows some sort of low-pass or bandpass filter that reduces the information content of the original signal to a level appropriate for the new sample rate.

In color image or video processing, this is often done to reduce the amount of data that needs to be transmitted or stored, taking advantage of the fact that the human visual system has poorer acuity for color information than it does for intensity. So, in a YCrCb representation, the Cr and Cb (color) components are filtered and then sub-sampled at 1/2 or even 1/4 the rate of the Y (intensity) component, with very little loss in terms of perceived image quality.

Answer 2

In image processing it often means chrominance subsampling, meaning that color information is in lesser resolution than the luminance (greyscale) data.

Sometimes, perhaps, it also could mean other things such as resampling to a lower resolution, for example to make it easier to calculate initial vectors in motion estimation.