Evaluation of clustering image using steady state genetic and hybrid K-harmonic clustering algorithms

Abstract EN

The goal of clustering is to determine the intrinsic grouping in a set of unlabeled data depending on some similarity measure (e.g.

Euclidean distance).In this paper a steady state genetic algorithm (SSGA) approach is used to cluster true color images.

After splitting the original images into red, green and blue components and displaying the image of each part, Steady State Genetic Algorithm (SSGA) is used to cluster the image to determine the number of clusters for the image by generating an initial population randomly and then applying the different operations of GA such as fitness function computation, selection, crossover, mutation and stopping condition.

In the Crossover stage 1X, PMX and UX methods used for crossover between two parents to produce a new child.

In addition to that another clustering method which combines kmean algorithms and k-harmonic mean algorithms are used.

The last clustering algorithm uses two functions to find the cluster centers for each image.

Finally root mean square error is used to find the difference between the clustering and original image.