| current international standard for image compression |
| lossy algorithm: the threshold of visible difference 20:1; ratio can be varied |
| designed for compression of full-color (24 bit) or gray-scale digitized images of "natural" scenes (continuous tone) |
| exploits the known limitations of the human eye (10,000 colors simultaneously) |
| DCT (Discrete Cosine Transformation) -- converts the spatial image representation (8x8 blocks) into frequency map |
| Huffman coding -- input is sliced into fixed units while the corresponding output comes in chunks of variable size |
| source available (Independent JPEG Group) |