JPEG-2000 (ITU T.800, ISO/IEC 15444) is a follow on to the original JPEG (ITU T.81, ISO/IEC 10918-1) standard published in 2000 for still image compression. JPEG-2000 provides an improvement in video compression efficiency by replacing the Discrete Cosine Transform (DCT) based algorithm with a new algorithm based on the Discrete Wavelet Transform (DWT). This compression efficiency gain can approach 20%.

 Since the DWT works on the entire image at one time and not just 8×8 blocks like the DCT, the blocking artifacts normally seen in highly compressed images are eliminated. However, highly compressed JPEG-2000 images may suffer from ringing artifacts. 

 JPEG-2000 decomposes the image into multiple resolution levels by alternating filter and subsample steps. Each level represents the image at a higher resolution. This allows a number of interesting features. Progressive transmission allows a low resolution image to be transmitted first, followed by progressively more resolution and detail. Similarly, only the lowest resolution images level can be sent as a thumbnail image. Also, the resulting image can easily be displayed in various resolutions.

 JPEG-2000 provides both support for both lossless and lossy compression modes. Use of a reversible DWT results in lossless compression and use of irreversible DWT will result in lossy compression. The reversible DWT is implemented in integer math which eliminates round-off errors.

 After the DWT is a quantization step which reduces the bit depth of the transform coefficients. This step is responsible for the loss in the lossy algorithm. Following quantization is a coding step that groups coefficients and efficiently codes them for transmission or storage. JPEG-2000 uses the Embedded Block Coding with Optimal Truncation (EBCOT) for this purpose.

 ISO/IEC 15444-1 defines the core coding system for JPEG-2000. Also defined is a motion JPEG-2000 (ISO/IEC 15444-3) which defines how to put together a sequence of still images to represent motion video.