A Survey of Image Segmentation Using K-means Clustering

In this survey paper we have tried to identify the necessity and advantages and disadvantages of different approaches of different researchers. We know the process of grouping objects or datasets with comparable data points is called clustering. Groups of items are as similar as they can be. Each group's points are as dissimilar as possible. One well- liked algorithm for classifying and dividing data is K-means clustering. It facilitates the separation of objects from their surroundings. According to K-means clustering, every feature point has a distinct location in space. In a multidimensional measurement space, a predetermined number of cluster centers are chosen at random by the fundamental K-means algorithm. The cluster with the closest mean vector is assigned to each pixel in the picture. This procedure continues until the cluster mean vector positions between iterations vary as little as possible. The K-means algorithm is significantly impacted by the initial starting points. Initial clusters are randomly generated by K-means. When these starting points are chosen close to the final solution, K-means can usually find the cluster center effectively. Otherwise, it could lead to poor clustering results. K-means clustering is one method for classifying data. The K-means function yields the cluster index for each observation following the division of the data into k distinct clusters.