Application of two fuzzy c-means clustering algorithms in segmenting the sonar image from a small underwater target into multi-regions

摘要

This paper expatiates on two image segmentation methods based on the improved fuzzy c-means (FCM) clustering algorithms. In the first method named as the method based on the two-dimensional histogram, each clustering sample is a two-dimensional vector structured the gray-level value of each pixel and mean value of each pixel neighborhood. In the second method named as the method considering pixel spatial information, clustering objective function, clustering center and membership function are modified by means of pixel spatial information. Both the first method and the second method take into account pixel spatial information by means of different techniques. The application to segmentation of a sonar image of a small underwater target shows that the first method is as effective as the second method.