
Decomposing a signal in time and frequency at once, for features that come and go.Wavelet analysis is a mathematical technique for breaking a signal into components that are localized in both time and frequency. Unlike the Fourier transform, which gives a global frequency spectrum and loses all time information, wavelets use small, finite 'waves' (the wavelets themselves) that are scaled and shifted to match features that appear and disappear. This makes it ideal for studying transient events, sharp edges, or signals with changing frequency content, such as seismic waves, heartbeats, or audio bursts.
The modern theory was developed in the 1980s by Jean Morlet and Alex Grossmann, building on earlier ideas in mathematics and signal processing. Ingrid Daubechies later constructed compactly supported orthogonal wavelets, making the method computationally practical.
A wavelet is a small oscillating function that is nonzero only over a short interval. The analysis works by comparing the signal to many scaled (stretched or compressed) and shifted (translated) versions of this wavelet. At each scale and position, a coefficient measures how well the wavelet matches the local shape of the signal. Large coefficients indicate strong similarity, revealing where and at what scale a feature occurs. The result is a two-dimensional map (scale vs. time) that shows how the signal's frequency content evolves.
Used in JPEG 2000 image compression, medical imaging (e.g., MRI denoising), earthquake detection, and analysis of gravitational wave signals.
The word 'wavelet' comes from the French 'ondelette', meaning 'small wave', and the first practical wavelets were developed by a geophysicist looking for oil.




