New methods in image compression using multi-level transforms and adaptive statistical encoDing
Abstract
The need to meet the demand for high quality digital images, with comparatively modest storage requirements, is driving the development of new image compression techniques. This demand has spurred new techniques based on time to frequency spatial transformation methods. At the core of these methods are a family of transformations built on basis sets called "wavelets." The wavelet transform permits an image to be represented in a substantially reduced space by transferring the energy of the image to a smaller set of coefficients. Although these techniques are lossy as the compression ratio rises, very adequate reconstructions can be made from surprisingly small sets of coefficients. This work explores the transformation process, storage of the representation and the application of these techniques to 24-bit color images. A working color image compression model is illustrated.