The Impact Analysis of the Effect of Noise in Natural Color Image Segmentation

In the segmentation process, an image is partitioned as a smaller group (cluster) of sub-images w.r.t some noteworthy attributes like color. The striking feature is that each and every component of image (pixels) in particular clusters have the inimitable features as compared to other cluster components. The image analysis always expects a desirable outcome from the segmentation process. The superfluous information (noise) by different sources tries to distort the original content. The quantity of unnecessary noise in the original image is inversely proportional to the quality of the outcome. The intended methodology employed median filtering, powerful nonlinear filter, to remove the unnecessary noisy pixels. In the proposed method, all the image components are sorted and the center value is calculated from them. The modified FCM clustering method is employed to huddle the noisy images. The image quality measures PSNR and computational cost is applied to measure the efficacy of the suggested methodology on natural images.