Signal Equalization
Multi-element detector alignment normalizes individual pixel gain and offset responses to ensure consistent spatial output across an optical array. Near infrared sorting systems employ non-uniformity correction to convert raw sensor outputs into quantitative reflectance data across passing textile items. The calibration process compensates for pixel response variations and non-uniform illumination fields produced by halogen or light-emitting diode light sources.
Standard procedures use reference target measurements to calculate gain and offset multipliers for each sensor element. The correction applies strictly to linear sensor responses and cannot repair saturated or dead detector pixels.
Two-Point Calibration
Optical calibration routines capture flat-field images at two distinct radiance levels to compute pixel multipliers. A dark reference captures baseline thermal currents, while a uniform white diffuse reflection standard defines individual gain factors. Applying non-uniformity correction ensures that identical fiber samples produce identical spectral signatures regardless of where they pass across the sensor field of view.
Uncalibrated arrays lead to misidentification of post-consumer garment blends near the edges of conveyor belts.
Sorting Application
High-speed textile sortation systems execute matrix calculations in hardware prior to spectral matching algorithms. Normalizing spatial optical variations enables precise quantification of cotton and polyester ratios in blended fabrics. Clean reflectance data prevents falsely classifying sound textiles as waste streams.
Drift Limit
Non-linear thermal expansion inside optical assemblies alters light distribution profiles over time. When illumination fields shift beyond the range of initial calibration arrays, stored calibration matrices produce signal artifacts that corrupt fiber spectral signatures.