L. Hertz et Rw. Schafer, POSTPROCESSING OF THRESHOLDED IMAGES TO MAXIMIZE EDGE COINCIDENCE, Journal of visual communication and image representation, 6(2), 1995, pp. 178-188
In many image processing applications, multilevel thresholding is a us
eful method for obtaining a simplified image that preserves the geomet
ric structures and spatial relationships found in the original graysca
le image. One way to ensure that they have been preserved is to ensure
that the edges of the original image are maintained in the thresholde
d image. This paper presents a new method for adjusting the value of e
ach global threshold to maximize the coincidence of the edges of the o
riginal and thresholded images. This paper also discusses two new meth
ods to improve the quality of images after thresholding. The result of
these algorithms is a final thresholded image that is a very good rep
resentation of the properties of the original scene. (C) 1995 Academic
Press, Inc.