Effect of Image Contrast and Resolution on Photointerpreter Target Detection and Identification*

Abstract

A psychophysical study was performed to determine the effects of image contrast and resolution on the extraction of information by a photointerpreter from images of Minuteman missile sites. Sixteen target films were selected, each containing a photograph of a different missile site. Sixteen degraded images of each site were processed, representing a 4×4 matrix of resolution and contrast combinations. Both of these image-quality variables were defined in terms of normalized MTF curves. Contrast was measured as the degree of exposure modulation at low spatial frequencies relative to that on the original images. Resolution was measured as the highest spatial frequency at which exposure modulation drops to that just perceivable by eye (about 0.022). The limiting ranges of these variables were preselected to extend from a relatively low (25%) to a relatively high (75%) probability of correct target-component identification. Each of 256 images was then shown to 16 experienced image interpreters. Accuracy and completeness of both target detection and identification was scored. The results indicate (a) that relationships between either contrast or resolution and either detection or identification completeness are roughly logarithmic over the ranges covered in this study; (b) that there is no regular relationship between accuracy and image quality; (c) that the effect of either image-quality variable (contrast or resolution) is diminished at the two higher levels of the other image-quality variable.