The Study Of Signal Decomposition Using Multi-Wavelet Based On DSP Builder

In China, the application of FPGA technology is no longer in the exploratory phase, with a more perfect development and design flow, but the difficulty of program using hardware language limits the development of FPGA technology. However, DSP Builder development environment proposed by Altera Co. presents a good solution to this problem. DSP Builder is a plug-in in Simulink, embedding in the form of toolbox. Components of its toolbox can be converted into hardware programming language by SignalCompiler modeling. By this way it will not only shorten the time of development process, but also reduce the cost of investment in hardware resources. CL multi-wavelet has a great advantage over the scalar wavelets in processing image edge and denoising, etc. Thus, this paper proposes that we achieve two parts:signal preprocessing and CL multi-wavelet decomposition based on DSP Builder which is used in signal processing, and image fusion theory with multi-wavelet as the basis. Firstly, the input image is dealt with repeated preprocessing and decomposed by multi-wavelet transform on MATLAB simulation tool, which shows that it outperforms scalar wavelets by comparing the entropy norm and who has a better decompositon effect in image processing. Then, this system is designed based on multi-wavelet and filtering bank theory, mainly including five parts:signal input section, pre-filtering subsystem, one-level multi-wavelet decomposition subsystem, two-level multi-wavelet decomposition subsystem, signal output section. The input of two-level decomposition subsystem is the low frequency signal of one-level decomposition subsystem. After completing the design of top-level modeling, the system is converted into VHDL language, also generates scripting files for third-party simulation. By analyzing the energy index, indicating that this system using CL multi-wavelet is superior to DB2 wavelet in signal decomposition. This system is not only the exploration of hardware implementation of multi-wavelet image fusion, but also the premise of achieving image fusion on SOPC.In the end, the paper summarizes the innovation of this project and prospects for multi-wavelet pre-filtering banks and how to realize multi-wavelet on DSP Builder.