Image Segmentation using FCM-Darwinian Particle Swarm Optimization

Segmentation of biomedical images is an essential requirement in image processing for assessment of different medical images i.e. microscopic, MRI and US. It can be a crucial step for decision support or can get the second opinion for medical expert for atypical cases. There are numerous segmentation methods available for different kind of images. An image segmentation method based on hybrid approach using Darwinian particle swarm optimizer and fuzzy C-means is implemented in this work for various medical and multimedia images. In the present work Darwinian particle swarm optimizer tries to solve the problems regarding the segmentation. The proposed method firstly initializes each of the particles present in the swarm with membership value of each pixel belonging to particular centroids with respect to fuzzy C-means and then optimizes the centroids values using Darwinian particle swarm optimizer. An efficient method for segmenting different areas and edges of various images is implemented in this work. For validating the output of proposed algorithm, it is compared with other segmentation techniques i.e FCM and FCM_PSO. Segmentation is evaluated on ground truth using various indexes. Finally, it is observed that the proposed technique turns out to be more consistent on segmenting the different medical and multimedia images.