JPEG
Verbosity reduction of YUV data (continued)
Each block of data is separated, and using the Discrete Cosine Transformation, is given a "DCT coefficient".The Discrete Cosine Transformation defines coefficients for the original data based on each picture block's frequency.
For example, if a picture block changes quickly (from black to white), its freqency is high; if it changes slowly (from black to gray to white), its frequency is low. Generally, the DCT coefficient of the picture block will be low. If amount of "0"s in the coeffient is high, then the compression ratio will also be high.