Scribd Logo
Upload

Welcome to Scribd!

  • 98 views

    Haar Wavelet

    Uploaded by

    cristi_bad
    The document discusses the discrete wavelet transform (DWT), including: 1) It describes a simple 2D image transform that decomposes an image into lower resolution components. 2) It explains the 1D Haar wavelet transform, how it can be represented as a matrix product, and how to reconstruct an image from its wavelet coefficients. 3) It discusses how the DWT can be used for image compression by removing high-frequency wavelet coefficients corresponding to fine image details.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Haar Wavelet

    Uploaded by

    cristi_bad
    98 views56 pages
    The document discusses the discrete wavelet transform (DWT), including: 1) It describes a simple 2D image transform that decomposes an image into lower resolution components. 2) It explains the 1D Haar wavelet transform, how it can be represented as a matrix product, and how to reconstruct an image from its wavelet coefficients. 3) It discusses how the DWT can be used for image compression by removing high-frequency wavelet coefficients corresponding to fine image details.

    Original Description:

    Transformata undisoara Haar

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses the discrete wavelet transform (DWT), including: 1) It describes a simple 2D image transform that decomposes an image into lower resolution components. 2) It explains the 1D Haar wavelet transform, how it can be represented as a matrix product, and how to reconstruct an image from its wavelet coefficients. 3) It discusses how the DWT can be used for image compression by removing high-frequency wavelet coefficients corresponding to fine image details.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    98 views56 pages

    Haar Wavelet

    Uploaded by

    cristi_bad
    The document discusses the discrete wavelet transform (DWT), including: 1) It describes a simple 2D image transform that decomposes an image into lower resolution components. 2) It explains the 1D Haar wavelet transform, how it can be represented as a matrix product, and how to reconstruct an image from its wavelet coefficients. 3) It discusses how the DWT can be used for image compression by removing high-frequency wavelet coefficients corresponding to fine image details.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 56

    Todays

    Topics
    9. The Discrete Wavelet Transform

    Topic 7:

    Discrete Wavelet Transform
    Wavelet transform vs. Laplacian pyramid
    Basic intui?on: a simple wavelet-like 2D transform
    The 1D Haar wavelet transform
    1D Haar wavelet transform as a matrix product
    Reconstruc?ng a 1D image from its wavelet coefs
    Wavelet-based image compression
    The 2D Haar wavelet transform

    Wavelet-Based Image Representa?ons

    The Laplacian Pyramid Representa?on

    Wavelet-Based Image Representa?ons

    Wavelet-Based Image Representa?ons

    Reminder: The Eigenface/PCA Image Basis

    Represen?ng Images by their PCA Basis

    The Discrete Wavelet Transform

    The Discrete Wavelet Transform

    Topic 7:

    Discrete Wavelet Transform
    Wavelet transform vs. Laplacian pyramid
    Basic intui?on: a simple wavelet-like 2D transform
    The 1D Haar wavelet transform
    1D Haar wavelet transform as a matrix product
    Reconstruc?ng a 1D image from its wavelet coefs
    Wavelet-based image compression
    The 2D Haar wavelet transform

    A Simple, Minimal 2-D Image Transform

    Transformed image (2Nx2N)

    Input image (2Nx2N)

    wavelet
    transform

    A Simple, Minimal 2-D Image Transform


    Step 1: Create 4 new images of size 2N-1x2N-1 as shown in
    gure
    Transformed image (2Nx2N)

    Input image (2Nx2N)

    Step 1

    A Simple, Minimal 2-D Image Transform


    Step 1: Create 4 new images of size 2N-1x2N-1 as shown in
    gure
    Transformed image (2Nx2N)

    Input image (2Nx2N)

    Step 1

    A Simple, Minimal 2-D Image Transform


    Step 1: Create 4 new images of size 2N-1x2N-1 as shown in
    gure
    Input image (2Nx2N)

    Transformed image (2Nx2N)

    Step 1

    A Simple, Minimal 2-D Image Transform


    Step 2: Recursively perform Step 1 for top-leW quadrant of
    result
    Transformed image (2Nx2N)

    Result of Step 1 (2N-1x2N-1)

    Average
    Image

    Step 2

    1 2
    3 4

    average

    A Simple, Minimal 2-D Image Transform


    Step 3: Recursion stops when average image is 1 pixel
    Transformed image (2Nx2N)

    Inver?bility of the Transforma?on

    1 2
    3 4

    Topic 7:

    Discrete Wavelet Transform
    Wavelet transform vs. Laplacian pyramid
    Basic intui?on: a simple wavelet-like 2D transform
    The 1D Haar wavelet transform
    1D Haar wavelet transform as a matrix product
    Reconstruc?ng a 1D image from its wavelet coefs
    Wavelet-based image compression
    The 2D Haar wavelet transform

    1D Haar Wavelet Transform: Recursive Deni?on

    1D Haar Wavelet Transform: Recursive Deni?on

    1D Haar Wavelet Transform: Recursive Deni?on

    1D Haar Wavelet Transform: Recursive Deni?on

    1D Haar Wavelet Transform: Recursive Deni?on

    Topic 7:

    Discrete Wavelet Transform
    Wavelet transform vs. Laplacian pyramid
    Basic intui?on: a simple wavelet-like 2D transform
    The 1D Haar wavelet transform
    1D Haar wavelet transform as a matrix product
    Reconstruc?ng a 1D image from its wavelet coefs
    Wavelet-based image compression
    The 2D Haar wavelet transform

    1D Haar Wavelet Transform as a Matrix Product

    1D Haar Wavelet Transform as a Matrix Product

    1D Haar Wavelet Transform as a Matrix Product

    1D Haar Wavelet Transform as a Matrix Product

    1D Haar Wavelet Transform as a Matrix Product

    1D Haar Wavelet Transform as a Matrix Product

    1D Haar Wavelet Transform as a Matrix Product

    The 1D Haar Wavelet Transform Matrix W

    Topic 7:

    Discrete Wavelet Transform
    Wavelet transform vs. Laplacian pyramid
    Basic intuiBon: a simple wavelet-like 2D transform
    The 1D Haar wavelet transform
    1D Haar wavelet transform as a matrix product
    ReconstrucBng a 1D image from its wavelet coefs
    Wavelet-based image compression
    The 2D Haar wavelet transform

    ReconstrucBng an Image from its Wavelet Coefs

    ReconstrucBng an Image from its Wavelet Coefs

    ReconstrucBng an Image from its Wavelet Coefs

    ReconstrucBng an Image from its Wavelet Coefs

    ReconstrucBng an Image from its Wavelet Coefs

    ReconstrucBng an Image from its Wavelet Coefs

    ReconstrucBng an Image from its Wavelet Coefs

    InterpreBng the Wavelet Coecients

    InterpreBng the Wavelet Coecients

    InterpreBng the Wavelet Coecients

    The Normalized Haar Wavelet Matrix

    The Normalized Haar Wavelet Coecients

    Topic 7:

    Discrete Wavelet Transform
    Wavelet transform vs. Laplacian pyramid
    Basic intuiBon: a simple wavelet-like 2D transform
    The 1D Haar wavelet transform
    1D Haar wavelet transform as a matrix product
    ReconstrucBng a 1D image from its wavelet coefs
    Wavelet-based image compression
    The 2D Haar wavelet transform

    Wavelet Compression Algorithm #1

    Wavelet Compression Algorithm #2

    Topic 7:

    Discrete Wavelet Transform
    Wavelet transform vs. Laplacian pyramid
    Basic intuiBon: a simple wavelet-like 2D transform
    The 1D Haar wavelet transform
    1D Haar wavelet transform as a matrix product
    ReconstrucBng a 1D image from its wavelet coefs
    Wavelet-based image compression
    The 2D Haar wavelet transform

    The 2D Haar Wavelet Transform

    The 2D Haar Wavelet Transform

    The 2-D Haar Wavelet Basis


    DeniBon of the rst few (coarsest scale) wavelet coecients of an
    image of dimensions of 2N x 2N

    +

    +
    Image
    Coef 3=
    Coef
    0 = Image
    +


    + 1 = Image
    +
    Image
    Coef 4=
    Coef

    +

    Image

    Coef 5=

    Image

    Coef 2 =

    A Simple, Minimal 2-D Image Transform

    The Haar 2-D Wavelet Transform


    The 2-D Haar Wavelet Transform corresponds to a modicaBon of
    this minimal recursive transform
    Input image

    Wavelet image
    representation (2Nx2N)

    (2Nx2N)

    Average Image:

    Step 1

    every pixel A is the


    average of pixels
    1,2,3,4

    Detail image

    Detail image

    Detail image

    1 2
    3 4

    detail
    coefficients

    InverBbility of the 2D Haar Transform


    We can recursively reconstruct the intensiBes of every 2x2 window
    from its average and detail coecients
    1

    2
    3 4

    1 2
    3 4

    2x2 pixel average

    d2

    d3

    d4

    d2

    d3

    d4

    You might also like