Uniformity from Lambertian optical sources such as integrating spheres is often trusted as absolute at levels of 98% (+/- 1%) or greater levels. In the progression of today’s sensors and imaging system technology that 98% uniformity level is good, but not good enough to truly optimize pixel-to-pixel and sensor image response. The demands from industry are often for “perfect” uniformity (100%) which is not physically possible, however, perfectly understood non-uniformity is possible. A barrier to this concept is that the definition and measurement equipment of uniformity measurements often need to be very specific to the optical prescription of the unit under test. Additionally, the resulting data are often a relativistic data set, assigned to an arbitrary reference, but not actually given an expression of uncertainty with a coverage factor. This paper discusses several optical measurement methods and numerical methods that can be used to quantify and express uniformity so that it has meaning to the optical systems that will be tested, and ultimately, that can be related to the Guide to the Expression of Uncertainty in Measurement (GUM) to provide an estimated uncertainty. The resulting measurements can then be used to realize very accurate flat field image corrections and sensor characterizations.
Sensor fusion and novel “multi-image” systems that have several different spectral ranges are proliferating in tactical and commercial applications. Calibrating these devices requires a variety of sources from quartz-tungsten halogen to blackbodies to more selectable band sources such as LEDs. Usually these sources are used independently in discrete spectral regions, but real reflective and emissive targets often have signatures that make combining these sources necessary if one is to emulate these real spectrums for testing in either image (collimator) or flood (sphere) configurations. A novel approach to combine LED and broadband emitters has been developed to effect stable, calibrated, traceable sources that can match real target spectral signatures.