Plato's Cave: Information Compression

Information Compression

Consider a digital image consisting of a lighter square (pixel value 1) on a dark background (pixel value 0). The most direct representation of this image in computer memory is as an exhaustive list of the raw pixel values, as shown below.

The fact that this image has multiple repeated values offers an opportunity for information compression. For example the same image can be compressed using a run-length encoding, for example 8(0) to represent a string of eight zero pixels.

Repetition in this compressed image offers the possibility of further compression, for example 2(8(0)) to represent two strings of eight zero pixels.

The symmetry in the original image allows for even further compression by defining "+" to represent four-fold symmetry, which means that only the upper-left quarter of the image need be encoded.

The point is that any kind of regularity or redundancy in the original image offers an opportunity for information compression.

Return to argument

Return to Steve Lehar