Effects of calibration–based non–uniformity correction methods on electro–optical performance of serial production thermal imaging systems

According to Planck's law of black body radiation, all objects with absolute temperature above 0 Kelvin emit electromagnetic radiation. Radiations emitted from objects in infrared wavelength can be interpreted by thermal imaging systems. Thus, thermal imaging systems are widely used in various applications across many fields such as military, medical, civilian, etc. Infrared focal plane arrays are the imaging sensors in thermal imaging systems that consists of a large number of infrared detectors arranged in a two–dimensional array and they convert the infrared radiation coming from scene into digital signals also known as wellfill levels. Due to several factors —including bot not limited to infrared detectors not being identical, defects in the coatings and alignment of optical units, the inhomogeneous distribution of incoming radiation onto the focal plane, environmental conditions— there exist some unwanted effects on the raw well–fill levels obtained by infrared focal plane arrays. When one or several of these effects are combined, they lead to a variance in the detectors' responsivity against infrared radiation, causing a non–uniformity effect on the raw well–fill levels obtained by the infrared focal plane array. Ideally, when an infrared focal plane array is positioned against a uniform radiation source, uniform viii well–fill levels would be expected across the detector array. However, the raw well–fill levels are not directly usable due to the non–uniformity effect. The responsivities of infrared detectors can be calibrated to uniform responsivities at specific points across the dynamic range using non–uniformity correction methods, in order for thermal imaging systems to function properly. The electro–optical performance of a thermal imaging system, with non–uniformity correction applied, can be indicated by residual non–uniformity. This measures the variance of well–fill levels from a uniform response when exposed to a uniform radiation source. The way of applying a non–uniformity correction method in thermal imaging systems is determined during the development phase, by taking into account the responsivity characteristics of a limited number of prototypes against a black body radiator that provides uniform radiation. Nevertheless, as the number of prototypes increases during the serial production phase, the responsivity and sensitivity of the prototypes against the black body radiator start to vary from one another. This variation may be caused by not just a single isolated source, but by multiple combined effects. These include, but are not limited to, the fact that subsystems such as optical, electro–optical, and electronic units are not exactly identical, due to manufacturing tolerances among many other reasons. The variance in responsivity and sensitivity among serial production prototypes may lead to them being calibrated at different points across the dynamic range and exhibiting different characteristics against infrared radiation, depending on the application procedure of non–uniformity correction methods. In this thesis, the raw well–fill data taken from 123 prototypes of a thermal imaging system that is in serial production since 2018 will be examined in terms of how they differs from one another across dynamic range and by using these well–fill data, different applications of different non–uniformity methods will be compared in terms of electro–optical performances. As a result, an automatic characterization and calibration setup that increases manufacturability will be presented which will apply non–uniformity correction on same points across dynamic range to thermal imaging system prototypes despite the responsivity and the sensitivity variances. Keywords: Thermal imaging systems, Infrared focal plane arrays, Non–uniformity correction, Serial production.