This chapter describes JPEG, a standard for lossy compression of still images based upon the discrete cosine transform (DCT). JPEG is rarely used directly in video, but it forms the basis of M-JPEG (used in desktop video editing) and DV compression. Also, JPEG techniques form the core of MPEG.

Motion-JPEG (M-JPEG) refers to the use of a JPEG-like algorithm to compress each field or frame in a sequence of video fields or frames. M-JPEG systems use the methods of JPEG, but rarely (if ever) conform to the ISO/IECJPEG standard. DV is a specific type of M-JPEG, which is well standardized; it is described in the following chapter, DV compression, on page 461. The I-frame-only variant of MPEG-2 is conceptually equivalent to M-JPEG, but again has a well-respected standard; see Chapter 40, on page 473.

The JPEG standard, cited in the margin, defines four modes: sequential, hierarchical, progressive, and lossless. The JPEG standard accommodates DCT coefficients having from 2 to 16 bits, and accommodates two different entropy coders (Huffman and arithmetic). Baseline refers to a defined subset of JPEG's sequential mode that is restricted to 8-bit coefficients and restricted to Huffman coding. Only baseline JPEG is commercially important. JPEG's other modes are mainly of academic interest, and won't be discussed here.