Glossary

9.3 Scaling and Wavelet Functions

4 min read•july 30, 2024

Scaling and wavelet functions are the building blocks of wavelet analysis. They work together to approximate signals at different scales and capture details. These functions have specific properties that make them effective for various applications like and signal denoising.

The multiresolution analysis framework connects scaling and wavelet functions. It creates a hierarchical structure where scaling functions provide coarse approximations, and wavelet functions capture the details between successive approximations. This approach allows for efficient signal representation and analysis.

Scaling and wavelet functions

Properties of scaling and wavelet functions

Top images from around the web for Properties of scaling and wavelet functions

  • Wavelet - Wikimedia Commons View original

    Is this image relevant?

  • Wavelet - Wikimedia Commons View original

    Is this image relevant?

1 of 2

Top images from around the web for Properties of scaling and wavelet functions

  • Wavelet - Wikimedia Commons View original

    Is this image relevant?

  • Wavelet - Wikimedia Commons View original

    Is this image relevant?

1 of 2
  • Scaling functions, denoted as ϕ(t)\phi(t), approximate functions at different scales or resolutions and serve as the building blocks of wavelet analysis
    • Must satisfy properties such as normalization, , and to ensure effectiveness in wavelet analysis
    • Designed to have specific properties like smoothness, symmetry, and vanishing moments depending on the application requirements (image compression, signal denoising)
  • Wavelet functions, denoted as ψ(t)\psi(t), are derived from scaling functions and capture the details or high-frequency components of a signal
    • Have a zero average value, allowing them to effectively represent fluctuations or changes in a signal
    • Also designed with properties like smoothness, symmetry, and vanishing moments based on the application

Multiresolution analysis (MRA) framework

  • Scaling and wavelet functions form a hierarchical structure known as a multiresolution analysis (MRA)
    • ϕ(t)\phi(t) approximates a signal at a coarse scale
    • Wavelet function ψ(t)\psi(t) captures the details or differences between successive approximations
  • Scaling function and wavelet function are related through the two-scale equation or refinement equation
    • Expresses the scaling function as a weighted sum of its scaled and shifted versions
    • Coefficients in the two-scale equation, known as scaling function coefficients or filter coefficients, determine the properties of the scaling and wavelet functions
  • In the MRA framework, the wavelet function is obtained from the scaling function by applying a high-pass filter, while the scaling function is associated with a low-pass filter

Relationship between scaling and wavelet functions

Two-scale equation and filter coefficients

  • Scaling function ϕ(t)\phi(t) and wavelet function ψ(t)\psi(t) are related through the two-scale equation or refinement equation
    • Scaling function is expressed as a weighted sum of its scaled and shifted versions: ϕ(t)=∑khkϕ(2t−k)\phi(t) = \sum_{k} h_k \phi(2t - k)
    • Coefficients hkh_k are the scaling function coefficients or filter coefficients
  • Filter coefficients determine the properties of the scaling and wavelet functions
    • Low-pass filter coefficients are associated with the scaling function
    • High-pass filter coefficients are associated with the wavelet function
  • Properties of the filter coefficients, such as their values and the number of non-zero coefficients, influence the characteristics of the scaling and wavelet functions (smoothness, support size)

Obtaining wavelet function from scaling function

  • Wavelet function ψ(t)\psi(t) can be obtained from the scaling function ϕ(t)\phi(t) using the high-pass filter coefficients gkg_k
    • Wavelet function is expressed as: ψ(t)=∑kgkϕ(2t−k)\psi(t) = \sum_{k} g_k \phi(2t - k)
    • High-pass filter coefficients gkg_k are related to the low-pass filter coefficients hkh_k through a specific relationship (quadrature mirror filter)
  • Scaling function acts as a low-pass filter, capturing the coarse approximation of the signal
  • Wavelet function acts as a high-pass filter, capturing the details or high-frequency components of the signal

Constructing wavelet basis functions

Dilation and translation operations

  • Wavelet basis functions are constructed from scaling functions through (scaling) and translation (shifting) operations
  • Dilation involves scaling the scaling function by a factor of 2j2^j, where jj is an integer representing the scale or resolution level
    • Allows wavelet basis functions to capture information at different scales
    • Example scales: j=0j=0 (original scale), j=1j=1 (half the original scale), j=2j=2 (quarter the original scale)
  • Translation involves shifting the dilated scaling function by integer multiples of the scale factor 2j2^j
    • Allows wavelet basis functions to cover the entire domain of the signal
    • Example translations: k=0k=0 (no shift), k=1k=1 (shift by 2j2^j), k=2k=2 (shift by 2⋅2j2 \cdot 2^j)

Wavelet basis functions and signal representation

  • Resulting wavelet basis functions, denoted as ψj,k(t)\psi_{j,k}(t), are indexed by the scale parameter jj and the translation parameter kk
    • Form a complete and orthonormal basis for representing signals
    • Expressed as: ψj,k(t)=2−j/2ψ(2−jt−k)\psi_{j,k}(t) = 2^{-j/2} \psi(2^{-j}t - k), where ψ(t)\psi(t) is the mother wavelet function derived from the scaling function
  • Signal can be decomposed into its wavelet coefficients by combining the wavelet basis functions at different scales and translations
    • Wavelet coefficients represent the contribution of each basis function to the signal
    • Allows for efficient representation, compression, and analysis of signals in the wavelet domain

Key Terms to Review (17)

Approximation Space: An approximation space is a mathematical framework that enables the representation and analysis of functions in a manner that captures essential features with a limited set of basis functions. This concept is crucial for signal representation, as it allows for the reduction of complexity while maintaining the fidelity of the original signal. The use of approximation spaces is fundamental when dealing with scaling and wavelet functions, as they provide the structure needed for multi-resolution analysis and efficient signal processing.
Compact Support: Compact support refers to a property of functions where the function is non-zero only within a compact subset of its domain, meaning it is zero outside of this bounded region. This characteristic is particularly useful in various areas like signal processing and wavelet theory, as it ensures that the function can be manipulated mathematically without affecting regions that are not of interest.
Daubechies Wavelet: The Daubechies wavelet is a family of wavelets that are used in signal processing and data compression, characterized by their compact support and the ability to provide a high level of smoothness with a minimal number of coefficients. These wavelets are designed to achieve orthonormality and are widely used for their effectiveness in multi-resolution analysis and feature extraction.
Detail coefficients: Detail coefficients are the values obtained from the wavelet transform that capture the high-frequency information of a signal, highlighting abrupt changes and transient features. These coefficients provide critical insights into the finer structures of the signal, enabling effective analysis in various contexts such as time-frequency localization, multi-resolution analysis, and biomedical signal processing.
Dilation: Dilation refers to the process of stretching or compressing a signal or function in time or frequency domains. It plays a critical role in scaling functions, particularly in wavelet analysis, where it allows for the adjustment of the resolution and detail of the signal representation. By manipulating dilation, one can analyze various aspects of signals at different scales, which is essential for understanding their structure and features.
Discrete Wavelet Transform (DWT): The Discrete Wavelet Transform (DWT) is a mathematical technique used to transform a discrete signal into its wavelet coefficients, enabling multi-resolution analysis. It addresses the limitations of traditional Fourier analysis by providing localized time and frequency information, allowing for better representation of non-stationary signals and images. DWT employs scaling and wavelet functions to analyze different frequency components at various resolutions, making it invaluable for tasks like image compression, watermarking, and biomedical signal analysis.
Fast Wavelet Transform (FWT): The Fast Wavelet Transform (FWT) is an efficient algorithm that computes the wavelet transform of a signal, significantly reducing the computational complexity compared to direct computation methods. By leveraging the properties of wavelet functions and their scaling relationships, the FWT allows for rapid analysis of signals across different frequencies and time scales. This efficiency is crucial in applications such as image processing, audio compression, and data compression, where quick processing of large datasets is necessary.
Haar wavelet: The Haar wavelet is a simple, step-like wavelet used in signal processing and image compression, characterized by its ability to represent data with sharp discontinuities. It is the first and simplest wavelet, making it foundational for understanding more complex wavelets and their applications in various analysis techniques.
Image compression: Image compression is the process of reducing the amount of data required to represent a digital image, allowing for efficient storage and transmission. It is essential for minimizing file sizes while maintaining acceptable visual quality, making it crucial for applications like digital photography, web graphics, and video streaming.
Multi-resolution decomposition: Multi-resolution decomposition is a powerful technique used in signal processing and analysis that breaks down a signal into various components at different resolutions. This allows for the examination of both the fine details and the broader trends within the signal, facilitating the analysis of non-stationary data. By employing scaling functions and wavelets, this method provides a framework for capturing and representing the essential characteristics of signals across multiple levels of detail.
Noise Reduction: Noise reduction refers to the techniques and processes used to eliminate or minimize unwanted disturbances in signals, enhancing the quality of the desired information. By employing various mathematical approaches and algorithms, noise reduction can improve the clarity and reliability of signals in different contexts such as images, audio, and other data forms.
Non-uniform scaling: Non-uniform scaling refers to the process of resizing an object or function in a way that does not maintain its original proportions, often altering its dimensions in different directions. This concept is crucial when dealing with wavelet functions, as it allows for flexible manipulation of signals and images, accommodating various applications such as image compression and feature extraction.
Orthogonality: Orthogonality refers to the concept of perpendicularity in a vector space, where two functions or signals are considered orthogonal if their inner product equals zero. This property is essential in signal processing and analysis as it enables the decomposition of signals into independent components, allowing for clearer analysis and representation.
Scaling Function: A scaling function is a mathematical function used in wavelet theory that helps define the way signals are represented at different resolutions. It acts as a basis for constructing multiresolution analysis, allowing for the decomposition of signals into various frequency components and enabling the representation of data at different scales.
Subband Coding: Subband coding is a technique used in signal processing that divides a signal into multiple frequency bands or subbands, allowing for efficient encoding and transmission of audio and video signals. This method takes advantage of the perceptual characteristics of human hearing and vision, leading to data reduction without significant loss of quality. By utilizing filter banks and discrete wavelet transforms, subband coding achieves better compression and noise resilience compared to traditional methods.
Uniform scaling: Uniform scaling refers to the process of resizing an object or function by the same factor in all dimensions, maintaining its proportions. This concept is essential in various mathematical fields, as it allows for consistent transformations that preserve the overall structure and relationships within a signal or function, particularly when dealing with wavelet functions and their multi-resolution analyses.
Wavelet transform: The wavelet transform is a mathematical technique that analyzes signals by breaking them down into smaller, localized wavelets, allowing for the representation of both time and frequency information simultaneously. This unique ability to capture transient features and varying frequencies makes it powerful for applications such as signal processing, image compression, and denoising.
© 2024 Fiveable Inc. All rights reserved.
AP® and SAT® are trademarks registered by the College Board, which is not affiliated with, and does not endorse this website.
Back
Practice QuizGlossary
Practice QuizGlossary
Next guide