JPEG is a still image compression algorithm developed by the Joint Photographic Expert Group, a joint committee of the ISO and the ITU-T. The algorithm was standardized in 1992 and published by the ITU-T as Recommendation T.81 and by the ISO/IEC as 10918-1.

 The JPEG algorithm is based on a Discrete Cosine Transform performed on 8×8 blocks of pixels. This transform, similar to the Fast Fourier Transform, converts the spatial information to frequency information. Quantization is then used to reduce the bit depth of and number of the frequency components. Finally, Huffman coding is used to further compress the remaining frequency components and prepare the bit stream for storage or transmission.

 JPEG supports a wide variety of image size, from very small (8×8) to extremely large (64Kx64K). It supports monochrome and color formats including RGB and YCbCr. JPEG also allows a variety of pixel component bit depths beyond the typical 8-bits per component. Component bit depths of 10, 12, 14 or 16 bits are often used for Infra-Red images.