基于提升算法的离散小波变换FPGA实现

FPGA Implemention of Discrete Wavelet Transform Based on Lifting Scheme

离散小波变换是当今许多图像处理和压缩技术的基础,并得到了广泛的应用.本文以4阶Daubechies小波为例阐述基于提升算法的离散小波变换的原理,并给出其硬件实现架构,然后进行仿真,将仿真结果与Matlab软件实现结果进行比较,结果表明硬件实现与软件实现基本一致,该硬件架构与基于传统的卷积方法实现相比,可以减小硬件实现面积,并利用插入流水线寄存器的方法,缩短关键路径,提高运算速度.

Discrete wavelet transform currently is the basis of digital graphic process and compression technology,and has been widely used. This paper introduces a FPGA implemention of D4 discrete wavelet transform based on lifting scheme. It mainly concerns the principle of lifting scheme, and the architecture of discrete wavelet transform based on lifting scheme. Then the architecture is simulated with Modelsim. Through comparing the result of simulation with the result implemented by Matlab, it shows that hard- ware implemention is consonant with software implemention. Meanwhile comparing this architecture with the architecture implemented based on convolution,it needs less area of hardware implemention. Through interleaving the pipelined registers, it can efficiently cut the critical path of the architecture based on lifting scheme, and then increase the speed of hardware implemention.