Image Segmentation via Iterative Fuzzy Clustering Based on Local Space-Frequency Multi-Feature Coherence Criteria

Fuzzy set theory has recently attracted much attention in the field of image classification, image understanding and image processing. One of the major topics in fuzzy image processing is the image classification problem. This paper presents a fast and accurate iterative fuzzy clustering (I.F.C.) method dynamically adapted to the classification process. This is used for high performance fuzzy segmentation which forms the basis for reliable image understanding. The proposed fuzzy segmentation scheme examines the image connectivity in the space and frequency domains. The detected fuzzy features are combined via a block synthesis and local correlation algorithmic procedure. Some results showing that the performance of the proposed I.F.C./clustering method is superior from that of the standard fuzzy c-means method are provided.