Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Signal Processing Course: Variational Regulariz...

Gabriel Peyré
January 01, 2012

Signal Processing Course: Variational Regularization

Gabriel Peyré

January 01, 2012
Tweet

More Decks by Gabriel Peyré

Other Decks in Research

Transcript

  1. J ( f ) = || f ||2 W 1,2

    = Z R2 ||r f ( x )||d x Sobolev semi-norm: Smooth and Cartoon Priors | f|2
  2. J ( f ) = || f ||2 W 1,2

    = Z R2 ||r f ( x )||d x Sobolev semi-norm: Total variation semi-norm: J ( f ) = || f ||TV = Z R2 ||r f ( x )||d x Smooth and Cartoon Priors | f| | f|2
  3. J ( f ) = || f ||2 W 1,2

    = Z R2 ||r f ( x )||d x Sobolev semi-norm: Total variation semi-norm: J ( f ) = || f ||TV = Z R2 ||r f ( x )||d x Smooth and Cartoon Priors | f| | f|2
  4. Function: ˜ f : x 2 R2 7! f (

    x ) 2 R ˜ f(x + ") = ˜ f(x) + hrf(x), "iR2 + O(||"||2 R2 ) r ˜ f ( x ) = ( @1 ˜ f ( x ) , @2 ˜ f ( x )) 2 R2 Gradient: Images vs. Functionals
  5. Function: ˜ f : x 2 R2 7! f (

    x ) 2 R Discrete image: f 2 RN , N = n2 f[i1, i2] = ˜ f(i1/n, i2/n) rf[i] ⇡ r ˜ f(i/n) ˜ f(x + ") = ˜ f(x) + hrf(x), "iR2 + O(||"||2 R2 ) r ˜ f ( x ) = ( @1 ˜ f ( x ) , @2 ˜ f ( x )) 2 R2 Gradient: Images vs. Functionals
  6. Function: ˜ f : x 2 R2 7! f (

    x ) 2 R Discrete image: f 2 RN , N = n2 f[i1, i2] = ˜ f(i1/n, i2/n) Functional: J : f 2 RN 7! J(f) 2 R J(f + ⌘) = J(f) + hrJ(f), ⌘iRN + O(||⌘||2 RN ) rf[i] ⇡ r ˜ f(i/n) ˜ f(x + ") = ˜ f(x) + hrf(x), "iR2 + O(||"||2 R2 ) r ˜ f ( x ) = ( @1 ˜ f ( x ) , @2 ˜ f ( x )) 2 R2 rJ : RN 7! RN Gradient: Images vs. Functionals
  7. Function: ˜ f : x 2 R2 7! f (

    x ) 2 R Discrete image: f 2 RN , N = n2 f[i1, i2] = ˜ f(i1/n, i2/n) Functional: J : f 2 RN 7! J(f) 2 R Sobolev: rJ(f) = (r⇤ r)f = f J(f) = 1 2 ||rf||2 J(f + ⌘) = J(f) + hrJ(f), ⌘iRN + O(||⌘||2 RN ) rf[i] ⇡ r ˜ f(i/n) ˜ f(x + ") = ˜ f(x) + hrf(x), "iR2 + O(||"||2 R2 ) r ˜ f ( x ) = ( @1 ˜ f ( x ) , @2 ˜ f ( x )) 2 R2 rJ : RN 7! RN Gradient: Images vs. Functionals
  8. rJ(f) = div ✓ rf ||rf|| ◆ If 8 n,

    rf[n] 6= 0, If 9n, rf [ n ] = 0, J not di↵erentiable at f . Total Variation Gradient ||rf|| rJ(f)
  9. rJ(f) = div ✓ rf ||rf|| ◆ If 8 n,

    rf[n] 6= 0, Sub-di↵erential: If 9n, rf [ n ] = 0, J not di↵erentiable at f . Cu = ↵ 2 R2⇥N \ (u[n] = 0) ) (↵[n] = u[n]/||u[n]||) @J(f) = { div(↵) ; ||↵[n]|| 6 1 and ↵ 2 Crf } Total Variation Gradient ||rf|| rJ(f)
  10. −10 −8 −6 −4 −2 0 2 4 6 8

    10 −2 0 2 4 6 8 10 12 −10 −8 −6 −4 −2 0 2 4 6 8 10 −2 0 2 4 6 8 10 12 p x 2 + " 2 | x | Regularized Total Variation ||u||" = p ||u||2 + "2 J"(f) = P n ||rf[n]||"
  11. −10 −8 −6 −4 −2 0 2 4 6 8

    10 −2 0 2 4 6 8 10 12 −10 −8 −6 −4 −2 0 2 4 6 8 10 −2 0 2 4 6 8 10 12 rJ"(f) = div ✓ rf ||rf||" ◆ p x 2 + " 2 | x | rJ" ⇠ /" when " ! +1 Regularized Total Variation ||u||" = p ||u||2 + "2 J"(f) = P n ||rf[n]||" rJ"(f)
  12. and 0 < ⌧ < 2/L, then f(k) k!+1 !

    f? a solution of min f J ( f ). If f is convex, C1 , rf is L -Lipschitz, Theorem: f(k+1) = f(k) ⌧k rJ(f(k)) f(0) is given. Gradient Descent
  13. and 0 < ⌧ < 2/L, then f(k) k!+1 !

    f? a solution of min f J ( f ). If f is convex, C1 , rf is L -Lipschitz, Theorem: f(k+1) = f(k) ⌧k rJ(f(k)) f(0) is given. Optimal step size: ⌧k = argmin ⌧2R+ J(f(k) ⌧rJ(f(k))) Proposition: One has hrJ(f(k+1)), rJ(f(k))i = 0 Gradient Descent
  14. Gradient Flows and PDE’s f(k+1) f(k) ⌧ = rJ(f(k)) Fixed

    step size ⌧k = ⌧ : Denote ft = f(k) for t = k⌧ , one obtains formally as ⌧ ! 0: 8 t > 0, @ft @t = rJ(ft) and f0 = f(0)
  15. J ( f ) = R ||r f ( x

    )||d x Sobolev flow: @ft @t = ft Heat equation: Explicit solution: Gradient Flows and PDE’s f(k+1) f(k) ⌧ = rJ(f(k)) Fixed step size ⌧k = ⌧ : Denote ft = f(k) for t = k⌧ , one obtains formally as ⌧ ! 0: 8 t > 0, @ft @t = rJ(ft) and f0 = f(0)
  16. @ft @t = rJ(ft) Noisy observations: y = f +

    w , w ⇠ N (0 , IdN ). and ft=0 = y Application: Denoising
  17. Optimal choice of t: minimize ||ft f|| ! not accessible

    in practice. SNR( ft, f ) = 20 log10 ✓ ||f ft || ||f|| ◆ Optimal Parameter Selection t t
  18. Sobolev prior: J(f) = 1 2 ||rf||2 f? = argmin

    f2RN E(f) = ||y f||2 + ||rf||2 (assuming 1 / 2 ker( )) Sobolev Regularization
  19. Sobolev prior: J(f) = 1 2 ||rf||2 f? = argmin

    f2RN E(f) = ||y f||2 + ||rf||2 rE(f?) = 0 () ( ⇤ )f? = ⇤y Proposition: ! Large scale linear system. (assuming 1 / 2 ker( )) Sobolev Regularization
  20. Sobolev prior: J(f) = 1 2 ||rf||2 f? = argmin

    f2RN E(f) = ||y f||2 + ||rf||2 rE(f?) = 0 () ( ⇤ )f? = ⇤y Proposition: ! Large scale linear system. Gradient descent: (assuming 1 / 2 ker( )) where ||A|| = max (A) ! Slow convergence. Sobolev Regularization Convergence: ⇥ < 2/||⇥ ⇥ ||
  21. Mask M, = diagi (1i2M ) Example: Inpainting Figure 3

    shows iterations of the algorithm 1 to solve the inpainting problem on a smooth image using a manifold prior with 2D linear patches, as defined in 16. This manifold together with the overlapping of the patches allow a smooth interpolation of the missing pixels. Measurements y Iter. #1 Iter. #3 Iter. #50 Fig. 3. Iterations of the inpainting algorithm on an uniformly regular image. 5 Manifold of Step Discontinuities In order to introduce some non-linearity in the manifold M, one needs to go log 10 (||f(k) f( )||/||f0 ||) k k E(f(k)) M ( f )[ i ] = ⇢ 0 if i 2 M, f [ i ] otherwise.
  22. Symmetric linear system: Conjugate Gradient Ax = b () min

    x 2Rn E( x ) = 1 2 h Ax, x i h x, b i
  23. Symmetric linear system: x (k+1) = argmin E( x )

    s.t. x x (k) 2 span(rE( x (0)) , . . . , rE( x (k))) Intuition: Conjugate Gradient Ax = b () min x 2Rn E( x ) = 1 2 h Ax, x i h x, b i Proposition: 8 ` < k, hrE( x k) , rE( x `)i = 0
  24. Symmetric linear system: Initialization: x (0) 2 RN , r

    (0) = b Ax (0) , p (0) = r (0) r (k) = hrE( x (k)) , d (k)i h Ad (k) , d (k)i d (k) = rE( x (k)) + || v (k)|| || v (k 1)||d (k 1) v (k) = rE( x (k)) = Ax (k) b x (k+1) = x (k) r (k) d (k) Iterations: x (k+1) = argmin E( x ) s.t. x x (k) 2 span(rE( x (0)) , . . . , rE( x (k))) Intuition: Conjugate Gradient Ax = b () min x 2Rn E( x ) = 1 2 h Ax, x i h x, b i Proposition: 8 ` < k, hrE( x k) , rE( x `)i = 0
  25. TV" regularization: (assuming 1 / 2 ker( )) f? =

    argmin f2RN E(f) = 1 2 || f y|| + J"(f) Total Variation Regularization ||u||" = p ||u||2 + "2 J"(f) = P n ||rf[n]||"
  26. TV" regularization: (assuming 1 / 2 ker( )) f(k+1) =

    f(k) ⌧k rE(f(k)) rE(f) = ⇤( f y) + rJ"(f) rJ"(f) = div ✓ rf ||rf||" ◆ Convergence: requires ⌧ ⇠ ". Gradient descent: f? = argmin f2RN E(f) = 1 2 || f y|| + J"(f) Total Variation Regularization ||u||" = p ||u||2 + "2 J"(f) = P n ||rf[n]||"
  27. TV" regularization: (assuming 1 / 2 ker( )) f(k+1) =

    f(k) ⌧k rE(f(k)) rE(f) = ⇤( f y) + rJ"(f) rJ"(f) = div ✓ rf ||rf||" ◆ Convergence: requires ⌧ ⇠ ". Gradient descent: f? = argmin f2RN E(f) = 1 2 || f y|| + J"(f) Newton descent: f(k+1) = f(k) H 1 k rE(f(k)) where Hk = @2E"(f(k)) Total Variation Regularization ||u||" = p ||u||2 + "2 J"(f) = P n ||rf[n]||"
  28. Noiseless problem: f? 2 argmin f J"(f) s.t. f 2

    H Contraint: H = {f ; f = y}. f(k+1) = ProjH ⇣ f(k) ⌧k rJ"( f(k) ) ⌘ ProjH( f ) = argmin g=y ||g f||2 = f + ⇤( ⇤ ) 1(y f) Inpainting: ProjH( f )[ i ] = ⇢ f [ i ] if i 2 M, y [ i ] otherwise. Projected gradient descent: f(k) k!+1 ! f? a solution of ( ? ). (?) Projected Gradient Descent Proposition: If rJ" is L-Lipschitz and 0 < ⌧k < 2/L,
  29. TV Priors: Non-quadratic better edge recovery. =) Optimization Variational regularization:

    () – Gradient descent. – Newton. – Projected gradient. – Conjugate gradient. Non-smooth optimization ? Conclusion Sobolev