Slides20

# Slides20 - Course 18.327 and 1.130 Course Wavelets and...

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Course 18.327 and 1.130 Course Wavelets and Filter Banks M-band wavelets: DFT filter banks and band cosine modulated filter banks. cosine Multiwavelets. M-channel Filter Banks channel • Used in communication e.g. DSL • 1 Scaling function, M-1 wavelets X(ω) H0(ω) ↓M ↑M F0(ω) H1(ω) ↓M ↑M F1(ω) M HM-1(ω) ↓M (↑2)(↓2)X(ω) = ½ [X(ω)+X(ω + π)] (↑M)(↓M)X(ω) = 1 M [∑ X(ω M-1 n=0 2πn +M )] ↑M FM-1(ω) ^ X(ω) 2 Perfect Reconstruction M-1 π 1 ∑ Fk(ω) M ∑ X(ω + 2Mn )Hk(ω n=0 k=0 1 M-1 2πn M-1 M-1 i.e. i.e. ∑ X(ω + M n=0 M π + 2Mn ) = e-iωlX(ω) π ) ∑ Fk(ω)Hk(ω + 2Mn ) = e-iωlX(ω) k=0 Matching terms on either side n=0 n≠0 ∑ Fk(ω)Hk(ω) = Me-iωl M-1 k=0 M-1 π ∑ Fk(ω)Hk(ω + 2Mn ) = 0 k=0 no distortion no aliasing no aliasing 3 e.g. M = 3 F0(ω)H0(ω) + F1(ω)H1(ω)+F2(ω)H2(ω) = 3e-iωl F0(ω)H0(ω + F0(ω)H0(ω + 2π 2π 2π ) + F1(ω)H1(ω + 3 ) + F2(ω)H2(ω + 3 ) 3 4π 4π 4π ) + F1(ω)H1(ω + 3 ) + F2(ω)H2(ω + 3 ) 3 Cast in matrix form Cast [F0(ω) F1(ω) F2(ω)] Hm(ω) = [3e-iωl 0 [F )] =0 =0 0] H0(ω) H0(ω+ 3 ) H0(ω+ 4π ) 3 Hm(ω) = H1(ω) H1(ω+ 23π ) H1(ω+ 43π ) H2(ω) H2(ω+ 23π ) H2(ω+ 43π ) 2π 4π 4 Polyphase Representation Polyphase Representation X(z) z-1 ↓M z-1 M z-1 ↓M ↓M ↓M X0(z) Y0(z) X1(z) Y1(z) X2(z) Hp(z) M XM-1(z) Y2(z) M YM-1(z) x[Mn] ↔ X0(z) = x[0] + z-1x[M] + z-2x[2M] + z-3x[3M] + … x[3M] x[Mn-1] ↔ X1(z) = x[–1] + z-1x[M-1] + z-2x[2M-1] + … 1] 1] x[Mn-2] ↔ X2(z) = x[–2] + z–1x[M-2] + z-2x[2M-2] + … 2] 2] M 5 To recover X(z) from X0(z), X1(z), X2(z), … (z), X(z) = ∑ zkXk(zM) X(z) M-1 k=0 k=0 Much more freedom than 2 channel case Much e.g. can have orthogonality & symmetry e.g. symmetry Consider Haar FB (M = 2) Consider Haar 11 Then Hp(z) = = F2 (2 pt DFT matrix) (2 1 -1 6 M-pt DFT matrix FM = 111 1 w w2 1 w2 w4 M 1) 1 wM-1 w2(M-1) 1 wM-1 w2(M-1) M w(M-1)(M-1) w= e-i 2π M 7 Suppose Hp(z) = FM … DFT DFT 14243 14243 … M pts M pts Y0(z) = ∑ Xn(z) (z) (z) M-1 n=0 ∑ wnXn(z) (z) n=0 M-1 M-1 ∑ x[-n] + M-1x[M-n] z-1 + … = ∑ n] n=0 n=0 M-1 M-1 ∑ wnx[–n] + ∑ wn-Mx[M-n] z-1 + … = n] n=0 n=0 Y1(z) = (z) M M-1 M-1 M-1 knX (z) = ∑ wknx[-n] + ∑ wk(n-M)x[M-n] z-1 + … Yk(z) = ∑ w n (z) (z) n] n=0 n=0 n=0 Terms in z-k are DFT coefficients of kth block of data. are So filter bank performs a block DFT. 8 Modulation followed by filtering x[n] x Z-1 (w0)-n x Z-1 (w)-n x M 2 -n (w ) Z-1 x (wM-1)-n p[n] ↓M p[n] ↓M p[n] ↓M p[n] ↓M For block DFT, p[n] = [1, 1, 1, … , 1] 0 M–1 • Can generalize by using other prototype filters. other • p[n] is called the prototype filter. modulation filtering If w–kn is replaced by ck,n from DCT ⇒ Cosine-modulated from 2 ck,n = √ M cos[(k + ½)(n + M + 1 ) π ] Filter Bank [(k )(n 2 2 M Filter M 9 Cosine Modulated Filter Bank (from type IV DCT) 2 hk[n] = p[n]√ M cos[(k + ½)(n + [(k M 2 1π + 2) M ] center it! “Double-shift shift p[n] chosen to be symmetric LPF. orthogonality” orthogonality Only p[n] needs to be designed. in M=2 case in Let L be the length of p[n]. Symmetry: P[L – 1 – n] = p[n] Symmetry: L=2M orthogonality: p[n]2 + p[n + M]2 = 1 L=2M orthogonality L=4M orthogonality: p[n]2+p[n+M]2+p[n+2M]2+p[n+3M]2 = 1 L=4M orthogonality p[n]p[n + 2M] + p[n+M]p[n + 3M] = 0 p[n]p[n 3M] Genus of the prototype filter. 10 10 Application to DSL copper wire f • assign more bits to lower frequency bands • orthogonal CMFB can undo the overlaps undo between channels between Multiwavelets Idea: extend the scalar refinement equation φ(t) = 2 ∑ h0[k]φ(2t – k) (t) (2t k into a vector refinement equation φ1(t) φ1(2t – k) (t) (2t =2∑ N-1 φ2 (t) (t) k=0 H0[k] 2x2 φ2 (2t – k) 11 11 e.g. Finite Elements e.g. Finite 1 φ(t) 0 1 0 1 1 ½ + 3 2 1 2 φ1(t) = φ1(2t) + ½φ1(2t – 1)+ ½φ2(2t) (t) (2t) (2t (t) 0 φ1(t) 1 1 ½ can use to represent piecewise linear can function but allows for representing function discontinuous function 1 φ2(t) 1 + 0 2 1 0 = ½ 1 1 φ2(t) = φ2(2t - 1) + ½φ2(2t) + ½φ1(2t-1) (t) (2t 1) (2t) (t) 0 1 (t) 12 12 ⇒ φ1 (t) (t) φ2 (t) (t) = 1 ½ φ1 (2t) (2t) ½ 0 φ1 (2t – 1) (2t 0 ½ φ2 (2t) ½ 1 φ2 (2t – 1) (2t f=1 f′=0 f=0 f′=0 f=0 f′=1 φ1(t) f=0 f′=0 φ2(t) 1442443 Finite Element Multiwavelets Finite Multiwavelets satisfy vector refinement equation can also come up with orthogonal multiwavelets. can multiwavelets 13 13 ...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online