Thresholding of a Digital Image by Free Terms.

The method of fuzzy segmentation of a digital image is generalized to the framework of L-fuzzy sets. Such an approach enables the use of various user-defined criteria for comparison of pairs of objects. These criteria are merged into a single mapping whose codomain is a poset. Then a free lattice structures are introduced and the procedure of L-fuzzy equivalence generation follows. The obtained L-fuzzy partition is then transformed into the resulting sets using the process of thresholding by terms from the free lattice. Discussion on applicability and computability of the results is presented and some examples are shown. Possible ways of improvements are suggested. [ABSTRACT FROM AUTHOR]