Wavelet Denoising Dsp Embedded In Near Real Time

In recent years, DSPs (Digital Signal Processors) have been developing rapidly because of its strong data processing ability and relatively cheap price.In this dissertation, a DSP-Embedded platform based on DSPs, was designed to prove the feasibility of the quasi real-time arithmetic of waveletIn the past twenty years, Wavelet Transform was a new subject developing quickly.The property of multi-resolution maked Wavelet Transform very valuable in many fields.According to the differencese of wavelet coefficients between information and noise transformed in different scale, the information could be mostly separated from the original signal.Firstly, comparing with other denoiosing methods, this thesis introduced the predominance of wavelet theory.And for fulfilling the requirement of real-time processe, it drew a DSP-Embedded project which was based on the quasi real-time arithmetic of wavelet theory.There was plenty of analysis for the quasi real-time and out-line wavelet arithmetic by MATLAB, otherwise the DSP-Embedded platform was designed for the quasi real-time wavelet arithmetic.On the platform, TMS320C6713 from TI Company was used to implement main function that processed floating-point data, and the C8051F320 was used as control module of the platform. In addition, for enhancing the quasi real-time arithmetic, parallel and serial communication, A/D function were all designed on the platform, and the platform could be compatible with USB2.0,PCI。With the quasi real-time arithmetic of wavelet theory, the platform reduced thenoise effectively form σ_orginal ,=1.0 to σ_denoised =0.65589.Otherwise, the effectwas very remarkable in the application of denoising the signal of liquid rate.Although the simulation and the application were limited by some objective conditions in this dissertation, it was feasible to implement the quasi real-time arithmetic of wavelet theory on the DSP-Embedded platform.