Research On The Methods Of Multispectral Infrared Imaging System Performance Characterization

With the increasing application of new technologies, processes and structures,multispectral imaging technology has become one of the most important developmenttrends in infrared imaging field. The utilization of new technologies leads to not onlyimprovement of system performance, but also the change of factors which affect thesystem performance. Hence, it is necessary to improve the traditional performancecharacterization method to satisfy the performance characterization and evaluation ofthe novel infrared imaging system (IRIS).This paper mainly involves research in the following several aspects:(1) For the requirement of triangle orientation discrimination (TOD) measurement,the relationship between correct discrimination probability of the human eye and visualthreshold signal-to-noise ratio (SNR) is studied through combing theoreticalcalculation with practical experiment. Based on the simulation images of trianglepatterns generated by an IRIS simulation model, relationship curves between these twovariables are yielded for different triangle angular subtenses (or sizes) with specifiedluminance, which is helpful not only to modify the existed single-band IRISperformance model but also to provide the basic data for the building of themultispectral IRIS performance model.(2) According to the working mechanism of multispectral imaging system, basedon the visually perceivable SNR, a TOD performance theoretical model of amultispectral staring IRIS with human vision is developed. Specifically, themathematical equations for predicting the TOD threshold of the system with differentfusion architecture are derived, with emphasis on the impacts of fusion on thethreshold. It yields a two-dimensional threshold surface in a three-dimensional graphwith the reciprocal angular subtense, the spectral difference, and the temperaturedifference as ordinates. Furthermore, a figure of merit Q related to the TODtemperature differential thresholds is introduced to analyze the relation of thediscrimination performance of multispectral IRIS to the size and the spectral differenceof test pattern. The preliminary validation with the experiment results suggests that thismodel can provide a reasonable prediction of the performance for a multispectral IRIS.(3) According to the difference between a multispectral IRIS and a single-bandIRIS in discrimining a field target, a new method for predicting the operating range of a multispectral IRIS is proposed by utilizing TOD performance model. The validationwith the experimental results with simulation images suggests that this approach canpredict the operating range of a multispectral IRIS efficiently.(4) Background clutter characterization for multispectral images and the effect ofbackground clutters on the detection performance of a multispectral IRIS are studiedpreliminarily. The application range of existed background clutter metrics isinvestigated through using multispectral images and statistical results of targetacquisition experiments. Then, the relationship between background clutter and thedetection performance of an IRIS is discussed. And the train of thought is proposed onhow to modify the detection performance prediction model of a multispectral IRIS byclutter metrics.