The integrated squared autocorrelation function of an image can be shown to represent a measure of pattern information or more commonly a measure of the amount sharpness and complexity of contour and contrast in the image. The autocorrelation function can thus provide a relative measure of image focus inasmuch as contour and contrast information is lost as an image goes out of focus. It is argued that the autocorrelation of the retinal image could be used as a feedback signal for control of visual accommodation and that such a mechanism is consistent with many well known phenomena of accommodation. In addition it is shown that the autocorrelation model can simulate quantitative data obtained from experiments on human accommodative behavior.