Some investigations on multiwavelet transform architecture for signal and image processing

Abstract

The development in the field of signal and image processing has motivated the researchers to design and implement various algorithms and mathematical tools in the recent past Converting a given data in time or space domain to other domains such as frequency with the help of transform technique is considered as a powerful mathematical tool by scientific community Among various transform techniques wavelet and multiwavelet transforms posses several desirable qualities They produce a transformed newlinesignal whose resolution say in time or frequency can be controlled by the user This flexibility made them a potential tool in signal processing applications which deal with a certain class of signals known as non stationary signals Multiwavelets can possess properties like symmetry orthogonality interpolating property and higher approximation order simultaneously Multiwavelets are application specific multiwavelet coefficients should be properly selected from the set of multiwavelet families In this research work a convolution based conventional systolic array and bit serial architecture for 1 D and 2 D DMWT is implemented at different levels of design abstraction newline