<title>Applications of recognition model to images of simulated vehicles</title>

Abstract

A computer model has been developed to predict the probability of recognition of particular shapes when viewed through a thermal imager employing either scanned or focal plane array detectors.This model is based on the results of a series of psychophysical trials during which human observers have considered over 120,000 images of shapes having a range of initial contrasts, and which have been degraded by various combinations of blurring and sampling. These computer generated images were presented to the observers in a random order and with a random degradation, using programmes to select images and display them on a computer monitor. After each presentation the observer decided which was the most likely shape to represent the image displayed on the screen. The responses collected have been used to calculate the human recognition probability of each image. A correlation has been found between the probability of recognition of any specified degraded shape and the relative contrast between the image of that shape, and the image of a similarly degraded circle of the same area. This model has been extended to include the effects of fixed pattern noise and applied to simplified images of cars and vans.

P-33: Identification of Simulated Targets as a Function of Target and Background Blur

“…Most of the perceptual research involving FLIR imagery seems to have been performed in the context of either testing particular visual models [8,9] or characterizing aspects of IR imagery that may be relevant to visual performance [1,2,4,5]. The data of Driggers et.…”