Nonuniformity correction and thermal drift compensation of thermal infrared camera

Abstract

In the last decade, a technology of thermal imagers based on new infrared detectors has been developed for both civil and military applications. These imagers implement miniaturized infrared detectors arranged in a matrix in the optical focal plane of the imager. FPA's technology links the detector matrix to specific electronics that allow detection and addressing at each pixel. This technology has allowed a rapid improvement in the performance of thermal imagers. Nevertheless, its use in thermographic measurement requires some metrological care. The first problem is the uniformity of the pixel response and the stability of this uniformity over time. The second problem is the compensation of the thermal drift. In this paper we present some practical solutions developed by CEDIP infrared systems to perform non-uniformity and thermal drift corrections. Performance and limitations are discussed.

Self-calibrating infrared thermometer for remote monitoring in cold climates

“…While some have suggested methods to predict this offset based on temperature measurements in the optical assembly [6] and predictions based on the rate of change of cold junction temperature [7], these methods will not produce accurate results when the thermal gradient is non-uniform or rotationally asymmetrical.…”

A scene-based nonuniformity correction algorithm based on fuzzy logic

“…The nonuniformity or fixed pattern noise (FPN) is a nonignorable drawback in the infrared focal plane array (IRFPA) [1,2]. In practical applications, it is necessary to apply a linear or nonlinear mapping to the response of each detector to make all detectors perform uniformly and reduce the FPN [1,3].…”

Bilateral filter-based adaptive nonuniformity correction for infrared focal-plane array systems

“…FPN is also characterized by a temporal drift, which is very slow and can be coherent ͓unstable bias voltage, focal-plane array ͑FPA͒ temperature control͔ or incoherent ͑generated within the individual FPA pixels͒. [1][2][3] In this context, NUC becomes necessary to eliminate the undesired effect of FPN and consequently to achieve higher quality infrared images. Generally, nonuniformity correction ͑NUC͒ techniques assume that the incident radiation is related to the readout signal by a "linear" relationship, which is uniquely identified by two parameters: the angular coefficient ͑gain͒ and the offset of the line.…”