Image segmentation is a necessary step in analysis of image. It is important to obtain high quality segments and in that choice of algorithm plays an important role. The clusters obtained from K-means Clustering, are improper as there is a double assignment of pixels. One cluster contains parts of other clusters, which is undesirable. This paper introduces a novel algorithm of Image segmentation, which modifies existing K-means clustering using Neutrosophic approach to remove the uncertainty in the assignment of pixels by changing the intensity of the pixel in such a manner that it gets assigned to only one cluster and thus making image suitable for segmentation. A number of experiments were conducted against different images. The quality of clusters and the clusters obtained are fine and smooth as has been proved by the proposed algorithms.