A continuous wavelet transform compares a signal with shifted and scaled versions of an analyzing wavelet. The coefficient array describes local structure across scales.
Mathematical principle
a denotes scale, b denotes position, and ψ* is the complex conjugate of the wavelet. This is an energy-normalized convention.
Input definitions
| Item | Meaning |
|---|---|
| Signal | Samples with a known sampling interval. |
| Wavelet | The analyzing family and its parameters. |
| Scales | The range and density of analyzed scales. |
Analysis choices
| Item | Meaning |
|---|---|
| Frequency mapping | Depends on the wavelet and sampling interval. |
| Normalization | Coefficient sizes depend on the convention. |
| Edges | Finite recordings introduce boundary effects. |
Interpretation
Larger scales correspond to more extended wavelets. Raw coefficient magnitudes should not be compared across different wavelet families without considering normalization. Record scale selection and boundary handling.
Scientific references
Related major models
From the principle to a scoped workflow
Record the wavelet family, scale grid, sampling interval, normalization and edge treatment together. A scalogram is a representation whose values need those definitions before comparison.
| Learning path | What to inspect and what it establishes |
|---|---|
| Inspect the measured signal first | Check timing, sample units, channels, gaps and the purpose of any preprocessing before choosing a transform. Evidence scope: This preparation lesson supplies no independent wavelet-calculation or event-detection acceptance. |
| Preserve method settings and output meaning | Keep the exact selected route and controls with the result. Record scale-to-frequency mapping and displayed quantity for the actual implementation. Evidence scope: A reference equation or a CWT label does not establish a particular wavelet family, algorithm, normalization or export contract in every product. |