Q1993Ingrid Daubechies constructed compactly supported orthonormal wavelets that made multiscale signal analysis practical for digital compression, denoising and reconstruction.
A Fourier analysis tells which frequencies exist, but wavelets can also preserve where a brief change or sharp edge occurs. Daubechies found finite, mathematically exact families that could be computed efficiently and reconstruct the original signal. Their descendants live in JPEG 2000, fingerprint storage, medical imaging and scientific data. Her abstraction succeeds because it is selective without being careless: it gives smooth areas less attention and spends detail where the world changes.


