Pattern recognition in the presence of visual noise.

Abstract

Recognition of a "target" pattern of dots against a background of "visual noise" was studied as a function of both the number of target dots and the number of noise dots. Data were obtained for each number of target dots from 2-9, using 192 different combinations of random target patterns embedded in random noise patterns varying from 1-8 dots. The Ss were 192 basic trainees at Lackland Air Force Base. The results indicate that increasing the complexity of the target pattern by increasing the number of elements improves recognition performance progressively. On the other hand, increasing the complexity of the visual noise produces a progressive decrement in recognition of the target. Increasing the number of elements in the target pattern beyond 7 or 8 seems to produce little if any further improvement in recognition performance. In general, recognition performance improves as the ratio of number of target dots to number of noise dots increases up to a target-to-noise ratio of about 31.