摘要
Decomposition and reconstruction of Mallat fast wavelet transformation （WT） is described. A fast algorithm, which can greatly decrease the processing burden and can be very easy for hardware implementation in real-time, is analyzed. The algorithm will no longer have the processing of decimation and interpolation of usual WT. The formulae of the decomposition and the reconstruction are given. Simulation results of the MEMS （micro-electro mechanical systems） gyroscope drift signal show that the algorithm spends much less processing time to finish the de-noising process than the usual WT. And the de-noising effect is the same. The fast algorithm has been implemented in a TMS320C6713 digital signal processor. The standard variance of the gyroscope static drift signal decreases from 78. 435 5 （°）/h to 36. 763 5 （°）/h. It takes 0. 014 ms to process all input data and can meet the real-time analysis of signal.
介绍了Mallat快速小波分解和重构算法,分析了一种可以大大降低运算负担,并且十分易于硬件实时实现的快速算法.该算法不再需要小波变换过程中的内插和抽取步骤,给出了相应的分解和重构过程的公式.对MEMS陀螺仪测量信号的仿真结果表明:算法只需更短的处理时间就可以完成去噪声过程,并且可以取得同样的去噪效果.在TMS320C6713芯片上实现了该算法,每个数据的处理时间只需0.014ms,静态漂移信号的标准差也从78.435 5(°) /h降到36.763 5(°) /h,完全可以满足信号实时处理的需要.
基金
The National High Technology Research and Devel-opment Program of China (863Program) (No2002AA812038)
