Architectures for wavelet transforms

A wide range of architectures for computing 1-D and 2-D DWT, and 1-D and 2-D CWT are presented. These architectures range from systolic arrays and parallel filters to SIMD arrays. The systolic array and the parallel filter architectures require an area that is independent of the length of the input sequence, and support single chip implementation. The SIMD architectures, on the other hand, are optimized for time, and have an area that is proportional to the size of the input.<
>