FY1988

The specific plan for this year is to complete and report a study measuring human visual abilities to detect objects in noise with the characteristics of air-to-air targets. The stimuli will be blurred points representing a distant aircraft on a near collision course moving at less than one pixel per frame. The primary variables will be:
1) Target Motion Uncertainty
A small effect of motion uncertainty will indicate the human visual systems ability to itself make simultaneous measurements of the likelihood of a target along multiple paths.
2) Temporal Integration
The interaction of the the temporal parameters of the human visual system and an image processing integration time should reveal the human system integration time for this task. Additional variables will include target velocity and target size.

The observer will be presented with two sequences one with a target and one without and asked to report which of the two sequences contained the target. Although cumbersome, the two interval forced choice procedure has the advantage of minimizing the effects of response bias on the measurement of detection thresholds. The level of the signal will be varied and a psychometric function estimated from the per cent correct as a function of the signal level. The masking noise will be pseudorandom noise using a machine independent algorithm for its generation, so that models for the human detection process will be able to be applied later to the identical noise sequences.

FY1999

The main focus of the second year will be on the development of models to adequately predict human detection performance in this situation. A second set of experiments will be designed to specifically distinguish among candidate models capable of predicting the results from the first year.

A second research concern of the second year will be the measurement of human performance in the presence of more sophisticated image processing algorithms, especially those attempting to characterize the spatial structure of the target from measurements over time along different potential paths.

FY2000

The third year of the program will be concerned with developing tools for the evaluation of combined image processing-human observer systems based on the models that will have been developed in the first two years.