Fpga-based Lifting Wavelet Transform

Lifting wavelet transform is a new wavelet transform method that can not use Fourier transform as major analysis tool. it can construct directly the wavelet in time-domain and frequency-domain. Lifting wavelet has been included in the JPEG2000 image compression standard. The general integral wavelet transform can be constructed by lifting arithmetic, design of circuit system of lifting wavelet transform is simplified greatly.In this dissertation, a research has been made on the lifting wavelet algorithm and its FPGA implementation architecture. 5/3 lifting wavelet system is realized in Altera's Stratix series chip.Firstly, the thesis introduces Multi-Resolution Analysis(MRA), double-scaling equations and lifting scheme. In accordance with the recursion algorithm for calculating lifting coefficients, An architecture of 5/3 wavelet decomposition and reconstruction is presented. Discuss the implementation of multiresolution decomposition and reconstruction of imageFocusing on lifting wavelet system design based on FPGA, the thesis analyzes the design of the wavelet transform, inverse transform, the overall control of implementation architecture within the main unit. In order to speed up wavelet transform, Multiplexer and displacement Calais instead of multiplication operation. Through transpose the memory address, row transformation can perform rank transformation. The wavelet coefficients is divided into four parts to store. Cacade the wavelet transform module, it can realize multi-level wavelet transform. At last, the architecture of inverse lifting wavelet transform is also proposed.Finally, the logic synthesis and timing analysis are implemented in Quartus II software with FPGA Stratix serial chip of Altera company.