Abstract
Derives a new compactly supported wavelet using the Daubechies approach. The construction of a "mother wavelet" is based on the notion of multiresolution analysis and is derived using the theory of compactly supported wavelet bases. The FIR filter related to this wavelet has 22 taps which leads to a regular wavelet with a high number of vanishing moments. The new wavelet and its dilated and shifted versions serve as a basis function for the measurable, square-integrable functions space L/sup 2/(R). To construct an orthonormal basis function for L/sup 2/(R/sup 2/), the authors simply take two one-dimensional bases and form the tensor product function. The new basis function is then implemented in a discrete form, and is used to decorrelate the data in an image, followed by a data compression scheme.<>