Pattern recognition is an emerging research area that studies the operation and design of systems that recognize patterns in data. Clustering is an essential and very frequently performed task in pattern recognition and data mining.Clustering refers to the process of grouping samples so that the samples are similar within each group. The groups are called clusters. k-means is one of the simplest unsupervised learning algorithms that solve the well known clustering problem. The procedure follows a simple and easy way to classify a given dataset of n points i x through a certain number of clusters fixed apriori. The difficulty in implementing k-means method for a large database is in determining the number of clusters which has to be randomly chosen. To overcome this difficulty, we propose a variation of the k-means algorithm, where the number of clusters ‘k’ can change dynamically depending on the data points and a threshold value given as an input. The proposed algorithm is applied in face recognition which is a very complex form of pattern recognition .It is used to verify whether a test face belongs to the database of faces and if so, identifies it.