In today's digital life, the segmentation of images is a very important issue. Each image contains a big amount of data. The internal relation between data of one image is nonlinear and ambiguous. There is a big uncertainty to find all segments of an image. Therefore, there is a big necessity to find a segmentation method for handling high uncertainty. In this paper, some of the previous works that have done on image segmentation have been improved and extended. A novel fuzzy c-means (FCM) is applied for the segmentation of images that inherently have high uncertainty and vagueness. A new method is used based on interval type-2 fuzzy sets, and the idea of reducing higher-order sets to lower order to capture the uncertainty of the images based on the decisiveness method. The higher peak signal to noise ratio (PSNR) value and the Jaccard similarity value for colour images show better segmentation results for having better performance and better effects on segmenting real world images. The results show that the proposed algorithm handles the segmentation of colour images better than the previous type-1 and type-2 FCM.