An error diffusion method suitable for producing an output image from an input image having a set of digitized continuous-tone pixels is disclosed. The method includes the steps of computing an image activity signal; computing a set of activity weights from the image activity signal; computing a filtered input value for a digitized continuous-tone input pixel responsive to the activity weights and computing a filtered output value responsive to the activity weights for each of the possible output levels. The method further includes selecting the output level in response to the filtered input value and the filtered output value for each of the possible output levels according to an error criterion; determining an error signal between the filtered input value and the filtered output value for the selected output level; and weighting the error signal and adjusting the filtered input values for nearby pixels which have not yet been processed.
展开▼
