Loading...
 

Anti-aliasing


In statistics, signal processing and related disciplines, aliasing is an effect that causes different continuous signals to become indistinguishable (or aliases of one another) when sampled. When this happens, the original signal cannot be uniquely reconstructed from the sampled signal.

Aliasing can take place either in time, temporal aliasing, or in space, spatial aliasing. It may give rise to moiré patterns (when the original image is finely textured) or jagged outlines (when the original has sharp contrasting edges, e.g. screen fonts). (Wikipedia).

Image
Aliased image.


Image
Plot showing aliasing of an undersampled 1D sinusoidal signal. The blue dots are the digital samples taken to record the red signal. Clearly, they are not enough to reconstruct the original signal: from then on, only the blue, wrong signal can be interpolated.


Avoiding aliasing

Anti-aliasing refers to the filters applied to an image before it is resampled (downscaled) to avoid Aliasing Artifacts. The filters remove high-frequency content from the original image reducing its Bandwidth. Therefore, the lower Sampling Density in the downscaled image is still above the Critical Sampling Distance.

Aliasing Artifacts are named after the fact that in undersampled images high spatial frequencies will return 'aliased' as low frequencies. Aliasing can easily occur in everyday pictures, and can cause disasters when periodic structures like the masonry in the following picture are present:


Strong aliasing artifacts caused by improper downscaling of this picture of the castle of St. Fargeau, France. The aliasing is most apparent in the masonry areas in the picture where high frequencies of the quasi periodic structure of the masonry are mapped to low frequencies.

A more subtle effect occurs in the areas with irregular textures: here, the aliasing effect gives rise to an overly coarse appearance of the grass, flowers and bushes.

Lastly, aliasing also gives rise to the staircase-like artifacts in the scaffolding on the left. This kind of artifact can be encountered in undersampled microscopic images of filament like structures.


Image


To the right the same picture, but now downscaled after applying an anti aliasing bandlimiting filter to the image first. Below a detail view of the top-right tower window.

Image



Image


Critical Sampling Distance
Nyquist Rate.