A robust thresholding algorithm for halftone dots

A robust and objective algorithm is proposed for the thresholding of halfto ne dot images during image processing. The algorithm is based upon detectio n of geometric features in the frequency histogram of grey level values. Im ages with a strong contrast between the halftone (lots and the imprinted ba ckground area tend to have a bimodal histogram, and the threshold is define d as the position of the minimum beni-cen the two peaks in the histogram. I mages with a weaker contrast between the halftone dots and the background t end to ha re an unimodal histogram. With Unimodal histograms it is necessar y to determine the point of inflexion on the histogram. The algorithm autom atically selects between these two approaches. The algorithm can also detec t those histograms where no threshold exists.