ImLab Scilab Function
Last update :

imkohonentrain - Kohonen networks method, training phase: second step of image segmentation using Kohonen networks clustering method.

Calling Sequence

[prototype] = imkohonentrain(learningdata,classnb[,iterationnb])

Parameters

Description

Thanks to learning data and following Kohonen networks clustering method, this function creates classes and returns their features in a prototype. The number K of classes is user defined.

Algorithm :

1. The nodes are initialized (one node for each class).

2. Choose one object.

3. Search the node which is the best match for this object. This node is the winner.

4. Refine the nodes that are excited by the winner.

Steps 2, 3 and 4 are repeated until all the objects have been chosen at least twice and at least 'iterationnb' have been done. The result prototype is formed with the nodes of the last iteration. Note that the distance used is the euclidean one.