A Reconfigurable Computing Engine for Wavelet Transforms

In the past a few years, wavelet transforms have become a hot topic of research. Discrete and continuous wavelet transforms have been widely used in signal and multimedia processing. Due to the high performance and flexibility of reconfigurable computing systems, it is very attractive to design a re configurable architecture for discrete and continuous wavelet transform of wide range of wavelet filters. In this paper, a unified computation framework for discrete and continuous wavelet transform based on lifting scheme and a reconfigurable architecture that includes reconfigurable lifting step arrays and reconfigurable address generator are proposed. The unified framework is the theory basis of this system. The step array is the computing core of this engine. And the address generator supports several memory scan pattern which is used to generate memory access addresses. In order to validate this architecture, an FPGA prototype is built based on Xilinx VirtexII FPGA to test the reconfiguration of 2-D discrete 5/3 and 9/7 transforms (defined in specification ofJPEG2000) and 2-D continuous Haar wavelet transform. Furthermore, a 3-level decomposition for a 512 times 512 grayscale image is performed and the results show that the decomposition can be finished within 12.16ms when running at 20MHz. It can be concluded that this design is applicable and scalable.