Spline wavelet image coding and synthesis for a VLSI based difference engine

Abstract

The efficiency of an image compression/synthesis system based on a spline multi-resolution analysis (MRA) is investigated. The proposed system uses a quadratic spline wavelet transform combined with minimum-mean squared error vector quantization to achieve image compression. Image synthesis is accomplished by utilizing the properties of the MRA and the architecture of a custom designed display processor, the Difference Engine. The latter is ideally suited to rendering images with polynomial intensity profiles, such as those generated by the proposed spline :V1RA. Based on these properties, an adaptive image synthesis system is developed which enables one to reduce the number of instruction cycles required to reproduce images compressed using the quadratic spline wavelet transform. This adaptive approach is computationally simple and fairly robust. In addition, there is little overhead involved in its implementation.