Framework for efficient optimal multilevel image thresholding

bstract. Image thresholding is a very common image processing peration, since almost all image processing schemes need some ort of separation of the pixels into different classes. In order to etermine the thresholds, most methods analyze the histogram of he image. The optimal thresholds are often found by either minimizng or maximizing an objective function with respect to the values of he thresholds. By defining two classes of objective functions for hich the optimal thresholds can be found by efficient algorithms, his paper provides a framework for determining the solution aproach for current and future multilevel thresholding algorithms. We how, for example, that the method proposed by Otsu and other ell-known methods have objective functions belonging to these lasses. By implementing the algorithms in ANSI C and comparing heir execution times, we can also make quantitative statements bout their performance. © 2009 SPIE and IS&T. DOI: 10.1117/1.3073891