Quantization of multiresolution transform coefficients for high compression of digital images

Blaine L Hagstrom, University of Nevada, Las Vegas


New developments in transformation theory have fueled interest in methods that employ transformation in the computation process. Theory from various disciplines including electrical engineering, physics, mathematics, and computer science have benefited from these advances. The greatest impact in computer science by these methods is in the area of image compression. Digital image compression is currently of high interest in computer science. The growing demand for images in computers has grown faster than the technology and thus solutions are sought. This work deals with the problem of quantization of resultant coefficients of the transforms in compression methods that perform transformation of the data. The digital image data transformations include quadrature mirror filtering, conjugate quadrature filtering, and wavelet methods. The process of transformation may be implemented in a reversible manner such that no change in the data is present. Quantization does not enjoy this luxury and implementation of a quantization scheme should be a careful and precise process. Various transformation processes are examined and the resultant data from the multiresolution sub-band coding process is targeted by quantization methods developed for compression of the data.