| Abstrakt: |
The performance of thermal imaging systems is markedly influenced by background effects. Two of those effects are discussed in this work: background clutter and its possible interference with target discrimination and background-scenery classification, which can be required to be performed simultaneously with target classification or to be used as an additional clue in the target-classification process. To integrate these effects into the existing model two concepts are suggested: discrimination of background details from targets based on classification at a level of orientation or recognition and an equivalent-target approach for modeling of background scenery. The design phase of a thermal imaging system is based on the transformation of the perceptual tasks into threshold thermal sensitivities at specific spatial frequencies. The background-augmented performance model defines multiple-threshold thermal sensitivity points, which are distributed throughout the frequency band of the system, in contrast to the single-point design of the existing model, which leads to an unbalanced optimization. By using the multiple threshold points generated by the background-augmented performance model, an optimal system design can be achieved. The design simultaneously covers aspects related to targets, generally located at the higher end of the spatial-frequency band of the system, and to background, which are usually found at the lower end of the spatial-frequency band. |
