A New Method For Predicting Field Discrimination Performance Of IR Imaging System

Field performance predictions are the discrimination capability of infrared imagingsensor for the typical target in the field environment, including range and probabilitypredictions under the different tasks levels: detection, recognition and identification.The impacts of atmospheric environment, a variety of physical effects of detector andclutter must be considered for field performance prediction. Considering complicatedcalculations, heavy workload and long-period field test, the traditional methods areadverse to field performance prediction. Therefore, a new method based on thermalimage is proposed to predict field performance in this paper.It is studied from the following three aspects. (1) Metrics and performancetheoretical models. Three primary theoretical models are introduced, the basic ideas offield performance accompanied with different metrics are described. (2) Laboratoryperformance curve testing. The SiTF, MRTD and TOD curves are determined,respectively through the laboratory testing. (3) The field performance prediction basedon thermal image. The target and background are first segmented from the thermalimages and their gray difference is estimated. Then, the testing curves in laboratory areused to implement field performance prediction with the Transfer Probability Function.Finally, comparisons with the field test data show that this field performance probabilityprediction method has a good accuracy.