Spatial Frequency

Spatial frequency is the frequency-domain analogue of spatial variation — the number of cycles per unit distance in a pattern or image, measured in cycles per meter or line pairs per millimeter. In optics and imaging, spatial frequency decomposes an image into sinusoidal gratings of different orientations and periodicities, analogous to how Fourier analysis decomposes a temporal signal into pure tones.

The concept is fundamental to understanding image resolution, contrast, and the modulation transfer function (MTF) of optical systems. A high spatial frequency corresponds to fine detail; a low spatial frequency corresponds to coarse structure. The maximum resolvable spatial frequency of an imaging system is set by its numerical aperture and wavelength — the diffraction limit — but aperture synthesis and superresolution techniques can recover spatial frequencies beyond this limit through prior knowledge or structured illumination.

In the visual system, neurons in early visual cortex are tuned to specific spatial frequencies and orientations, a discovery that earned Torsten Wiesel and David Hubel the Nobel Prize. The brain performs a form of spatial frequency analysis on the retinal image, decomposing the visual world into a multiscale, multi-orientation representation. This biological fact has inspired image compression standards like JPEG and JPEG 2000, which exploit the same frequency-domain decomposition.

The spatial frequency domain is where the physics of imaging meets the information theory of sampling. Any theory of vision, photography, or remote sensing that ignores spatial frequencies is not incomplete — it is blind.