Also found in: Dictionary, Thesaurus.


The elementary building blocks in a mathematical tool for analyzing functions. The functions can be very diverse; examples are solutions of a differential equation, and one- and two-dimensional signals. The tool itself, the wavelet transform, is the result of a synthesis of ideas from many different fields, ranging from pure mathematics to quantum physics and electrical engineering.

In many practical applications, it is desirable to extract frequency information from a signal—in particular, which frequencies are present and their respective importance. An example is the decomposition into spectral lines in spectroscopy. The tool that is generally used to achieve this is the Fourier transform. Many applications, however, concern nonstationary signals, in which the makeup of the different frequency components is constantly shifting. An example is music, where this shifting nature has been recognized for centuries by the standard notation, which tells a musician which note (frequency information) to play when and how long (time information). For signals of this nature, a time-frequency representation is needed.

There exist many different mathematical tools leading to a time-frequency representation of a given signal, each with its own strengths and weaknesses. The wavelet transform is such a time-frequency analysis tool. Its strength lies in its ability to deal well with transient high-frequency phenomena, such as sudden peaks or discontinuities, as well as with the smoother portions of the signal. (An example is a crack in the sound from a damaged record, or the attack at the start of a music note.) The wavelet transform is less well adapted to harmonically oscillating parts in the signal, for which Fourier-type methods are more indicated.

Applications of wavelets include various forms of data compression (such as for images and fingerprints), data analysis (nuclear magnetic resonance, radar, seismograms, and sound), and numerical analysis (fast solvers for partial differential equations).

wavelet compression

A lossy compression method used for color images and video. Instead of compressing small blocks of 8x8 pixels (64 bits) as in JPEG and MPEG, the wavelet algorithms compress the entire image with ratios of up to 300:1 for color and 50:1 for gray scale.

Wavelet compression also supports nonuniform compression, where specified parts of the image can be compressed more than others. There are several proprietary methods based on wavelet mathematics, which are available in products from companies such as Summus, Ltd. (www.summus.com) and Algo Vision LuraTech GmbH (www.luratech.com). See lossy compression and JPEG 2000.

Amazing Compression Ratios
This picture was compressed at 156:1 with InfinOp's Lightning Strike software. The original file was more than a million bytes and compressed down to 7.5K. This image is amazingly good especially considering that it contains only 128 colors. (Image courtesy of InfinOp, Inc.).
References in periodicals archive ?
Algorithms employing wavelets uses arithmetic code as it allows fractional number of bits in encoding.
In results obtained in Table 1, the emphasis is on the algorithms employed rather than the role played by wavelets. It could be seen that Peak to Signal Noise Ratio (PSNR) and Mean Square Error (MSE) values remain almost same for different frequency of compression in different methods.
Majhi, "Brain MR image classification using two-dimensional discrete wavelet transform and AdaBoost with random forests," Neurocomputing, vol.
Acharya et al., "Application of wavelet techniques for cancer diagnosis using ultrasound images: A Review," Computers in Biology and Medicine, vol.
Haar wavelet transform is a kind of discrete wavelet transform.
Burras, "Wavelet transforms and filter banks", Wavelets: A Tutorial in Theory and Applications, vol.
The quantitative properties of any wavelet method strongly depends on the used wavelet basis, namely on its condition number, the length of the support of the wavelets, the number of vanishing wavelet moments, and the smoothness of the basis functions.
The voxel patterns precede the wavelets; thus, this important property is kept in wavelets.
According to Figure 6 and Table 3, all the wavelets above are insensitive to the change point at t=0.3s because the data amplitude difference between the end point of oblique envelope signal and the initial point of exponential envelope signal is very small.
There are several well-known families of orthogonal wavelets. An incomplete list includes Harr, Meyer family, Daubechies family, Coiflet family, and Symmlet family [28].
A wavelet is a wave-like oscillation with an amplitude that begins at zero, increases, and then decreases back to zero.