Main Article Content
Background: Fuzzy c means (FCM) clustering is an unsupervised clustering technique used for image segmentation. It is widely used in the image segmentation process due to its simplicity and efficiency. But it also has some weaknesses which need to be addressed and overcome for quality results. Optimization techniques are used to overcome these drawbacks.
Objectives: This paper presented a comprehensive study of bio-inspired optimization-based Fuzzy c-means clustering which tries to overcome these drawbacks of FCM and improve the FCM segmentation results. The study also shows the importance of bio-inspired optimization-based FCM in image segmentation. It includes almost all the important bio-inspired optimization techniques that are used to improve or hybridize FCM for better image segmentation results. An experimental setup is performed using five PSO, ABC, FA, GA and TLBO evolutionary algorithms to segment an image. It provides insight into determining which optimization techniques are best done with FCM.
Methods: The FCM based on five PSO, ABC, FA, GA and TLBO evolutionary algorithms is implemented on Berkeley’s test images using Matlab 2013b. The experimental setup is performed for each algorithm separately and results are recorded.
Results: The results for each of these algorithms are compared based on convergence and computation time. From the results, it is found that the PSO-FCM outperforms other optimization algorithms.
Conclusions: from this study it is concluded that FCM with optimization techniques can effectively overcome the standard FCCM drawbacks. The experiments show that the PSO-FCM outperforms other optimization techniques in terms of convergence and computation time.