南京理工大学数字信号处理英文课件 Lecture22.ppt

296 * -----Application: Signal Analysis Using the DFT ( chapter 11 ) ? General setting ? Effect of windowing and frequency sampling ☆ smears or broadens the impulses in the theoretical Fourier representation ? exact frequency is less sharply defined ☆ reduces the resolution Lecture 22 Discrete Fourier Transform ( V ) Parameters: ? window length N and shape ? DFT size L ? sampling period T (neglect aliasing here ) Block diagram of continuous-time signal analysis by using DFT A. Principle C/D ? L-point DFT V[k] w[n] T x[n] v[n] x c[t] B. Example: sinusoidal signals Let x c ( t ) = cos ( ?0 t ) + A cos (? 1 t ) Then : ? x[n] = cos (ω0 n ) + A cos (ω1 n ), with ωi = T ? i C. Windowing: resolution and leakage ( I ) Example: Rectangular window ? Attention: ? Position of the frequency in DTFT and DFT. The peak of spectral in DTFT -----resolution V(e j?) ≠0, for most k -----leakage Windowing: resolution and leakage ( II ) Windowing: resolution and leakage ( III ) Windowing: resolution and leakage ( IV ) Windowing: resolution and leakage ( III ) Resolution-----width of the mainlobe of window function( W(e j?)); Degree of leakage-----relative amplitude of the mainlobe and the sidelobe of window function( W(e j?)); Increasing L( length of window) cannot increase the resolution ? Increasing N( length of DFT)can increase the resolution D. Frequency matched to sampling points Frequency matched to sampling points ( II ) N.B: if Hamming window ! Attention: different spectrum with different window E. Increasing the resolution Not by increasing L, the DFT size: Increasing the Resolution ( II ) Not by increasing L, the DFT size: Increasing the Resolution ( III ) Not by increasing L, the DFT size: Increasing the Resolution ( IV ) By increasing N , the window size: * * *