| Image Science is the process by which the
measurements of imaging system primitives are used to construct
and instantiate metrics that are predictive of the customer's
assessment of quality. Image system primitives are determined
from measurements of test targets such as an ISO resolution
target for digital still cameras and color test targets. |
Sharpness
The ability to render fine detail can be determined by the
imaging system's ability to render a sharp edge or line. The
system primitive obtained from the is measurement is know as the
Modulation Transfer Function. By combining this with the human
eye's response, the observer centric metric of Subject Quality
Factor (SQF) is determined. SQF is linearly correlated with
perceived quality. |
 |
Memory Colors
 Using targets of known color stimuli, color reproduction errors
can be computed in a visually uniform color space such as CIELab.
These errors, expressed in terms of deltaE can be correlated with
perceptual image quality. |
Noise
 In a truly uniform target area, there is no information. Any
deviation in the value of the image data is therefore evidence
of unwanted image noise. The granularity metric is obtained from
the imaging primitive measurement of Noise Power Spectrum, which
measures the spatial distribution of the sign fluctuations. When
combined with the human eye response, this yields a metric that
is linearly correlated with perceived quality. |
Tone Scale
The system primitive corresponding to the tone scale is derived
from the characteristic curve of the imaging system - the
relation between stimulus and response - expressed in a visually
uniform color space such as CIELab. |
 Customer
Image Quality
Image Integration provides the tools and metrics to enable the
prediction of customer perceived quality from these perceptually
based imaging primitives. |
