Fuzzy partition of two-dimensional histogram and its application to thresholding

This paper proposes a thresholding approach by performing a fuzzy partition on a two-dimensional (2D) histogram of the image. The novel 2D fuzzy partition method and membership assignment are based on fuzzy relationship of two fuzzy sets characterized by a set of the best parameters (a, b, c), which makes the image have the maximum fuzzy entropy. The threshold is selected as the crossover point of the fuzzy region Ca, cl. The preprocessed moment arrays are employed to reduce the computation time of image entropy from 0(8N(2)) to 0(2N) The proposed approach has been tested on various images, and the results have demonstrated that the proposed 2D fuzzy approach outperforms the 2D nonfuzzy approach and the one-dimensional(1D) fuzzy partition approach. (C) 1999 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.