The quality of a photointerpreter's anno- tation of an image may be displayed by writing maps in which the pixels are shaded according to the number of times a pixel was seen to be on a feature. The map of grey tones is more valuable, if, for each grey level, a measure of quality is determined, which is the probability that a feature exists in a pixel of a given grey tone. From repeated annotations (at least four) of the same image by a single observer, it is possible to formulate a perception model which describes stochastic processes in the observer._x000D_
From that model, it is possible to assign estimates of data quality for each observer to each grey tone. For two observers on the same scene, the 'mean' method of quantifying the grey levels is to assign a value equal to the mean of the ii,:lities for each of the two observers separately. For this example, the grey tones arr thus quantified at probability levels of 0.70, 0.80, 0.90 and 0.95.