Aliasing degrades imagery and affects visual task performance. This chapter describes a model that predicts the effect of sampling on target identification. Aliasing is treated as noise. The combined effect of aliasing and detector noise degrades the system contrast threshold function (CTFsys). The degraded (elevated) CTFsys lowers the TTP resolution. The effect of aliasing on PID is predicted by the TTP metric.
Details on the thermal and reflective models are described in Chapters 9 and 10, respectively. Those chapters provide details on calculating CTFsys for different types of imagers. This chapter explains modeling concepts.
Aliasing acts like noise because of the imaging task. We are interested in quantifying expected or average performance over many object identification attempts. At all ranges, the imager is presented with a diverse target set. The objects are placed randomly in the field of view. At each range, the task is repeated many times. The size and placement of spatial features varies from target to target. There are many targets in many different scenes. Aliasing acts like noise because of the combination of multiple targets, target diversity, random target placement, and the random sample phase of various target details.
The aliased signal is different from detector noise in two ways. First, aliasing disappears as the target contrast disappears. The amplitude of aliasing depends on target contrast. Second, the image corruption due to aliasing gets worse with increased range. This is because sampling is constant in angle space, and targets become poorly sampled as range increases.
Total noise is the quadrature sum of detector noise and aliasing. In order to sum the two noises, they must be properly scaled. Section 8.1 explains how signal and detector noise are spatially normalized in the thermal and reflective models. Section 8.2 describes the different temporal treatments of noise in detectivity versus photon-counting models. As far as the eye is concerned, noise is noise. However, the disparate treatment of noise in the two types of models results in different calibration constants.