Multiwavelets And Its Application To Image Compression

Methods for image compression have become the subject of much study over the past decade, especially those based on wavelet transform have been developed quickly lately. The newly developed multiwavelets offer more design options and hence can combine all the desirable properties such as orthogonality, symmetry, compact support and high approximation order. But perfect multiwavelet in theory will still confront with many problems in practice. More degrees of freedom add more difficulties for the construction of multiwavelet system; The filter of multiwavelet is matrix filter, so it is necessary to pre-process the scalar signal before filtering to obtain vector-valued signal; Because of the special subband structure of multiwavelt transform, the quantization method based on scalar wavelet should not be used for image compression directly.According to these problems mentioned above, this thesis gives some research on the construction of multiwavelet systems and prefilters and the application of multiwavelets to image compression. It includes the construction of biorthogonal multiwavelet with high approximation order;The construction of a kind of prefilter according to the given symmetric-antisymmetric multiwavelet;A comprehensive analysis of multiwavelet properties and its effect on image compression;Three quantization methods modified according to multiwavelet based on embedded zerotree wavelet(EZW) coding and analysis of their applicability on different multiwavelet; A much larger selection of images with different characteristics are test to give a comparison of multiwavelet and scalar wavelets with the intention of dertermining which is the best wavelet base and which is the best quatization method for each image. The experimental results prove that the application of multiwavelet to image compression has many advantages and great potentials.