JPEG 2000

JPEG 2000 is a new ISO/ITU-T standard for coding static images. (December 1999). It is intended for low-bit operations and subjective quality image. The file extension is '.j2k'
Advantages of use :

High compression (0.25 bit per point in the case of grayscale image)
Loss and loss-free compression in one source code
Smooth quality and scaling ability, without downloading the whole file. The main advantage is maintaining the data stream.
Support of very small images: JPEG is limited to 64kx64k. JPEG 2000 up to 2^32-1.
One decompression architecture
Error resilience for transmission in noisy environments, such as wireless and the Internet
Computer generated imagery
Compound documents
Region of Interest coding
Improved compression techniques to accomodate richer content and higher resolutions
Metadata mechanisms for incorporating additional non-image data as part of the file

JPEG 2000 can have up to 256 channels of information; in comparison to JPEG, which is limited to RGB data.

Effectiveness of compression

According to the results achieved, JPEG 2000 is by 20 % better than JPEG.

This factor is given :

Storing loss and loss-free compressions
oooooooo?    Multiple-component images
     Static and dynamic Region-of-interest
     Error resilience
     Spatial and quality scalability


Further new standards are the following: JPEG-LS, and MPEG-4 VTC. We can compare them against the older standards the JPEG and the PNG.


This very popular ISO/ITU-T standard supports only loss compression! The algorithm is based on dividing an image into 8x8 blocks, and these are subsequently transformed by DCT. The transformed blocks are quantized with a uniform quantizer, then arranged using the zig-zag method, and coded using the Huffman coding. In the quantization process, each of the 64 DCT coefficients is firmly determined for the whole image. The DC coefficients from all blocks are coded separately

The loss-free version is called L-JPEG. It is, however, based on a completely different principle than the JPEG. It uses the prediction scheme. The prediction is based on three adjacent points and seven different predictions.


The MPEG-4 Visual Texture Coding (VTC) is an algorithm used in the MPEG-4 for the compression of textures and static images. It is based on the discrete wavelet transformation (DWT), scalar quantization, 0-tree coding and arithmetic coding. The quantization is scalar, and has three types: single (SQ), multiple (MQ), and bi-level (BQ). With the SQ, every wavelet coefficient is quantized once. With the MQ, it is used the common quantizator. The BQ in essence is a SQ, merely the information is sent in bit levels. The MPEG-4's disadvantage is that it does not support a loss-free compression.


This is the last ISO/ITU-T standard for loss-free coding of static images. It supports ' virtually-loss-free' coding. It is based on adaptive prediction, context modeling, and the Golombovam coding. The virtually-loss-free coding has a fixed maximum amount of error points in comparison to the original


The Portable Network Graphics (PNG) is a W3C specification for coding static images. It is based on the prediction scheme and entropy coding. Prediction is based on a comparison and predicting of neighboring points. Entropy uses the Deflate algorithm based on the LZ77. The PNG is a loss-free coding.

JPEG 2000

The JPEG 2000 is based on the discrete wavelet transformation (DWT), scalar quantization, context modeling, and arithmetic coding