Histogram based fuzzy Kohonen clustering network for image segmentation

Abstract

In this study, a Fuzzy Kohonen Clustering Networks (FKCN) algorithm is proposed for image segmentation. The large image sizes are in general required for various types of imaging applications. For large amounts of data, computation of Fuzzy Kohonen Clustering Networks (FKCN) algorithm for iterative operation will take a long time. Therefore, it has been attempted to make the FKCN algorithm realistically useful for image segmentation. In this work, a proposed FKCN algorithm will keep the advantage of using this plan. But instead of using image space, a gray level function of image was used for all imaging data. A new algorithm was developed for applications of various images. Good results were taken in a short time for image segmentation.

Modified Fuzzy Kohonen Clustering Network for image segmentation

“…Bezdek, Pal and Taso [6] considered this approach. , they extended the ideas of FCM and Kohonen neural network to a new family of algorithms called Fuzzy Kohonen Clustering Network (FKCN).…”

“…One popular fuzzy technique involves using fuzzy c-means(FCM) for image segmentation [2].The neural network approaches have been introduced which rely on a neural network architecture for image segmentation, also the hybrid of neural network with fuzzy is used in image segmentation [3] [4].The hybrid learning scheme was a first attempt to merge fuzzy clustering and feature maps by Huntsberger and Ajjmarangsee [5 ] .They attempted to establish a connection between feature maps and fuzzy clustering by modifying the learning rule proposed by Kohonen for the SOFM . Hamdi used Fuzzy Kohonen neural network for image segmentation, in [6]. The theoretical foundations of this paper are based on theoretical foundations of Fuzzy Kohonen Neural Network in [7].…”

A fuzzy self‐organizing map algorithm for biological pattern recognition

“…Their ‘Kohonen‐type FCM’ algorithm, which is called fuzzy Kohonen clustering networks (FKCN), however, was only utilized to analyse clusters within Iris data set. Only later did other researchers use the method for different tasks including image segmentation (Atmaca et al ., ), texture segmentation (Liu et al ., ), and biological macromolecule image classification (Pascual et al ., ). In another relevant study, Vuorimaa () substituted the neurons of SOM with fuzzy rules to obtain more precise clustering.…”

Color image segmentation using a self-organizing map algorithm

“…Many approaches to color segmentation have been developed, including fuzzy c-means ͑FCM͒, 2-4 region growing and merging, 5-7 region growing and edge information, [8][9][10][11] the Markov random field ͑MRF͒, 12-15 the neural network, 16 -18 histogram-based methods, [19][20][21][22][23][24] and the physically based method. 25 For a more detailed review of gray and color image segmentation methods, the reader is referred to a general review given in Ref.…”