Chapter 2 Continous Wavelet Transform (CWT).ppt

The properties of Daubechies Wavelet: Highest number of vanishing moments for a given support width. Orthogonal, Biorthogonal Compact Support Support width 2N-1 No Symmetry 1. Daubechies Wavelet Daubechies Wavelet is wrote for short “dbN”, N = 2 ~ 10, is the rank. When N = 1, db1 is Haar wavelet. db8 Wavelet function [phi,psi,xval]=wavefun(‘dbn’, iter) */43 db8 Scaling function 2. Symlets wavelet It is the improving of Daubechies wavelet. It is wrote for short “symN”, N = 2~8. [phi,psi,xval]=wavefun(‘symn’, iter) The properties of Symlets wavelet: highest number of vanishing moments for a given support width Orthogonal, Biorthogonal Compact Support (width 2N-1) Near Symmetry */43 sym8 Wavelet function sym8 Scaling function 3．Meyer Wavelet The properties: regular orthogonal wavelet Orthogonal, Biorthogonal No Compactly Supported, Effective support=[-8 8] Symmetry [phi,psi,xval]=wavefun(meyr, 5); Wavelet function Scaling function */43 Summary */43 Pulse function 2.6 CWT of some classical signals */43 Sinusoidal function */43 Two sinusoidal function with different frequencies */43 f1=sin(0.01*pi*(t/1000).*t); t=1:2000; c=cwt(f1,1:198,mexh); Unstable signals */43 */43 练习：用C语言或Matlab编程实现对某一维函数的连续小波变换。 Email: ymq_hw@ */43 Homework1 [phi1,psi1,Xval]=wavefun(db8,9); psi2=psi1(1:2:end); psi4=psi1(1:4:end); figure,plot(Xval,psi1) ylim([-2.4,2]) figure,plot(Xval,sqrt(2)*[psi2,zeros(1,3840)]) ylim([-2.4,2]) figure,plot(Xval,2*[psi4,zeros(1,3840+1920)]) ylim([-2.4,2]) * Onset 开始时刻 * Distinguish * 小波消失矩的作用 我们通常用的函数dbn中的n就是这个小波函数的消失矩;消失矩越大,它的支撑长度就越大,通常是支撑长度不少于2*n-1的;消失矩越大,对应的滤波器越平坦,而且小波函数的振荡很强.光滑函数在利用小波展开后的零点越多,也就是说小波的消失矩的大小,决定了小波逼近光滑信号的能力.这一点也可以用来进行图像压缩.越大的消失矩将使高频系数越小,小波分解后的图像能量也就很集中,压缩比例就越高.通常我们都愿意采用消失矩较高的小波函数. 我们可以对一个信号,采用不同的消失矩的小波函数来分解,就可以更加感性的了解它.. 由图中我们可以看出消失矩增大时,它的高频分量中的零越来越多。 因此，总结起来有下面的结论： 1）消失矩大，小波函数光滑； 2）支撑长度长； 3）压缩率大； 4）信号分解后，高频分量少，低频分量多； 5）消失矩大，小波函数光滑，支撑长度长。但其傅立叶变换恰好相反，即消失矩小的话，其频域支撑长度长且光滑性好； 6）消失矩大，滤波器的长度越长。 /zjfldp/item/8f5488e92d6470 %%%%%%%