Where does sharpness collide?

In imaging science, sharpness is the perceived crispness of an image, governed by how abruptly tone changes across an edge, and it is quantified objectively as acutance, derived from the steepness of the measured edge profile. Higgins and Jones's 1952 The nature and evaluation of the sharpness of photographic images showed that subjective sharpness tracks the mean-square density gradient at edges and introduced acutance as its physical correlate. The referent is an edge-rendering property of an imaging system, distinct from resolution, measured from edge-spread or modulation-transfer data. It shares only its name with the hearing and forecasting meanings.[1]

Unit

The steepness of the tone transition across an image edge (acutance)

Recovery means

Not applicable

Measured by

Quantified by acutance from edge-gradient measurements and by the system's modulation transfer function

In psychoacoustics, sharpness is a sensation of the timbre of a steady sound that rises as its spectral energy shifts toward high frequencies, largely independent of overall loudness. Zwicker and Fastl's Psychoacoustics: Facts and models formalizes it as a weighted first moment of the specific-loudness pattern across critical bands, expressed in a unit called the acum. The referent is a perceptual attribute of sound quality, used in noise and product-sound engineering, computed from the loudness spectrum rather than from any image or probability distribution.[2]

Unit

The acum - a high-frequency-weighted centroid of specific loudness across critical bands

Recovery means

Not applicable

Measured by

Computed from the specific-loudness pattern as a high-frequency-weighted centroid, in acum

In probabilistic forecasting, sharpness is a property of the forecasts alone: the degree to which predictive distributions are concentrated, so that narrower, more committal predictions are sharper. Gneiting, Balabdaoui, and Raftery's 2007 Probabilistic forecasts, calibration and sharpness frames the aim as maximizing sharpness subject to calibration, measured by the width of prediction intervals or the spread of the predictive density. The referent is a concentration property of a forecast distribution, unrelated to an image edge or an auditory timbre beyond the shared word.[3]

Unit

The spread or interval width of a predictive distribution

Recovery means

Not applicable

Measured by

Assessed by average prediction-interval width or the dispersion of predictive densities, independent of the outcomes