Digital imaging of still images and video has become widespread in consumer, industrial, military, and scientific markets in large part because of compression -- shrinking the original image file to a fraction of its original size by reducing redundancy within the image. The compressed file not only consumes significantly less memory space, but it now can be easily transmitted electronically over transmission media with limited bandwidth. All compression algorithms accomplish this in one way or another, but it would be even more useful if the image's compressed codestream could be structured in such a way so as to provide simple and flexible mechanisms to access different representations of the original image.
Image files that are compressed using traditional compression algorithms such as JPEG or MPEG are only useful once they are decoded; in their compressed state, the images are locked in a state in which nothing can be extracted or scaled. Only when the file is decoded can any further processing be done. Image or video encoding in the JPEG or MPEG format is analogous to a TV program being recorded on an analog tape (a form of compression in the analog domain). By itself, the recorded tape is completely useless and only becomes useful when it is put into a video player. Only when the tape is being played back on a player can the video be reprocessed or scaled. In short, digital imaging, in the case of JPEG or MPEG, has not advanced beyond the capabilities of analog imaging.
There is, however, an algorithm that generates a compressed codestream that retains the ability to easily extract different representations of the image for various end applications without the need to recompress the codestream. This capability allows the user to move beyond the accepted paradigm of what can be done with a compressed image (i.e., basically nothing). In some applications, it would even make sense to have the image compressed for transmission even if there is enough bandwidth, since the compressed image is in a more manageable, robust format than its original state. This compression algorithm is called JPEG2000.
Developed more than six years ago by the Joint Photographic Experts Group, JPEG2000 has gradually become more prevalent in a variety of industries such as surveillance, medical, satellite, and more recently, professional video (including Digital Cinema) and consumer markets. Supporting both lossless and lossy compression, JPEG2000, unlike JPEG or MPEG, is based on wavelets -- a mathematical function that separates a signal into different frequency bands, and then enables analysis of each frequency band depending on the resolution or the scale (See "An Introduction to Wavelets," Amara Graps, IEEE Computational Science and Engineering, Summer 1995, Volume 2, Number 2.) In addition to this inherent flexibility, JPEG2000 also provides a very significant improvement in compression efficiency compared to the baseline JPEG algorithm.
NEXT: Wavelet Encoding, Scalability