DSP第十七次课(第九章).ppt

Chapter 9 DFT/FFT Algorithms DFT : discrete Fourier transform FFT : fast Fourier transform (FFT is the fast implementation of DFT) FFT have three major uses in DSP : a) the numerical computation of the frequency spectrum of a signal b) the efficient implementation of convolution by the FFT c) the coding of waveforms 9.1 Frequency Resolution and Windowing 问题：用数字信号处理的方法分析模拟信号x(t)的频谱 用我们前面学过的知识解决这个问题： 第一步：对x(t)进行采样得到x(n) 第二步：对x(n)作DTFT得到 ，那么怎样由采样后的信号频谱 得到模拟信号频谱 呢？ 利用采样后信号频谱与模拟信号频谱的关系 结论：当采样后频谱不出现混叠时，在Nyquist间隔内的采样后频谱就是模拟信号频谱，当混叠不严重时，可用Nyquist间隔内的采样后频谱近似模拟信号频谱。 (Because digitally we can only compute , it is essential that Eq.(9.1.1) be satisfied , or that the extra terms in Eq.(9.1.2) remains small over the Nyquist interval ,which happens when falls off sufficiently fast with f ) 在实际中用DTFT分析模拟信号频谱还会遇到两个问题。 第一个问题： Even though is the closest approximation to that we can achieve by DSP ,it is still not computable because generally it requires an infinite number of samples x(nT) , . To make it computable ,we must a second approximation to , keeping only a finite number of samples , say, x(nT), . This time-windowing process is illustrated in Fig.9.1.1 (9.1.7) As the length L of the data windows increases ,the windowed signal becomes a better approximation of x(n), and thus , a better approximation of ( Driving the precise connection of the windowed spectrum to the unwindowed one of Eq.(9.1.7) . ) L=20 L=20 (The mainlobe peak at DC dominates the spectrum , because w(n) is essentially a DC signal ,except when it cut off at its endpoints . The higher frequency components that have “leaked” away from DC and lie under the sidelobes represent the sharp transition of w(n) at the endpoints .) Two Important Parameter of Windowed Spectrum 1)Mainlobe